JP7517774B2 - Velocity-based control in aperiodically updated controller, method for controlling a process, process controller - Google Patents

Description

This patent relates to methods and system compensation for providing rate-based control in process control systems using non-periodic control or slow feedback process variable communication, and more specifically to devices and methods configured to robustly control a process when implementing control when receiving process variable feedback at a slow rate compared to the process dynamics.

A process control system, such as a distributed or scalable process control system such as those used in chemical, petroleum, or other processes, typically includes one or more process controllers communicatively coupled to each other, to at least one host or operator workstation, and to one or more field devices via an analog bus, a digital bus, or a combined analog and digital bus. The field devices, which may be, for example, valves, valve positioners, switches, and transmitters (e.g., temperature, pressure, and flow sensors), perform functions within the process, such as opening or closing valves and measuring process parameters. The process controller receives signals indicative of process measurements and/or other information made by the field devices and uses this information to implement control routines to generate control signals that are sent over the bus to the field devices to control the operation of the process. Information from the field devices and controllers can typically be utilized by one or more applications executed by the operator workstation, allowing an operator to perform any desired function with respect to the process, such as ascertaining the current state of the process, modifying the operation of the process, etc.

Some process control systems, such as the DeltaV® system sold by Emerson Process Management, use function blocks or groups of function blocks, referred to as modules, located in a controller or in different field devices to perform control and/or monitoring operations. In these cases, the controller or other device may include and execute one or more function blocks or modules, each of which receives inputs from and/or provides outputs to other function blocks (either in the same device or in different devices) to perform some process operation, such as measuring or sensing a process parameter, monitoring a device, controlling a device, or performing a control operation, such as implementing a proportional-integral-derivative (PID) control routine. The different function blocks and modules in a process control system are generally configured to communicate with each other (e.g., over a bus) to form one or more process control loops.

A process controller is typically programmed to execute a different algorithm, subroutine, or control loop (all of which are control routines) for each of a number of different loops defined for or contained within a process, such as flow control loops, temperature control loops, pressure control loops, etc. Generally speaking, each such control loop includes one or more input blocks, such as an analog input (AI) function block, a single-output control block, such as a proportional-integral-derivative (PID) or fuzzy logic control block, and an output block, such as an analog output (AO) function block. The control routines, and the function blocks that implement such routines, are configured according to a number of control techniques, including PID control, fuzzy logic control, and model-based techniques such as Smith Predictor or Model Predictive Control (MPC).

To support the execution of routines, a typical industrial or process plant has a centralized control room communicatively connected to one or more process controllers and process I/O subsystems, which are in turn connected to one or more field devices. Traditionally, analog field devices are connected to the controller by a two-wire or four-wire current loop for both signal transmission and power delivery. Analog field devices (e.g., sensors or transmitters) that transmit signals to the control room modulate the current through the current loop so that the current is proportional to the sensed process variable. Meanwhile, analog field devices that perform actions under the control of the control room are controlled by the magnitude of the current through the loop.

With the increasing volume of data transfer, one particularly important aspect of process control system design involves the manner in which field devices are communicatively coupled to each other, to controllers within the process control system or process plant, and to other systems or devices. In general, the various communication channels, links, and pathways that enable the field devices to function within the process control system are commonly collectively referred to as an input/output (I/O) communication network.

The communication network topology and physical connections or paths used to implement an I/O communication network can significantly affect the robustness or health of field device communications, especially when the network is subject to adverse environmental factors or harsh conditions. These factors and conditions can impair the health of communications between one or more field devices, controllers, and the like. Communications between controllers and field devices are particularly sensitive to any such disruptions, since monitoring applications or control routines typically require periodic updates of process variables for each iteration of the routine. Thus, impaired control communications can result in reduced efficiency and/or profitability of the process control system, and excessive wear or damage to equipment, as well as any number of potentially harmful failures.

To ensure robust communication, I/O communication networks used in process control systems have historically been hardwired. Unfortunately, hardwired networks introduce numerous complexities, challenges, and limitations. For example, the quality of hardwired networks can deteriorate over time. Furthermore, hardwired I/O communication networks are typically expensive to install, especially when the I/O communication networks are associated with large industrial plants or facilities that are distributed over a large area, such as refineries or chemical plants that consume many acres of land. The required long wiring installations typically involve significant labor, materials, and costs, and may introduce signal degradation resulting from wiring impedance and electromagnetic interference. For these and other reasons, hardwired I/O communication networks are generally difficult to reconfigure, modify, or update.

The use of wireless I/O communication networks has been proposed to alleviate some of the problems associated with hardwired I/O communication networks. See, for example, U.S. Patent Application Publication No. 2003/0043052, entitled "Apparatus for Providing Redundant Wireless Access to Field Devices in a Field Communication System," the entire disclosure of which is expressly incorporated herein by reference.
"Distributed Control System" discloses a system that utilizes wireless communications to augment or supplement the use of hardwired communications.

Reliance on wireless communications for control-related transmissions has traditionally been limited due to reliability concerns, among other things. As explained above, modern monitoring applications and process control rely on reliable data communication between controllers and field devices to achieve optimal levels of control. Furthermore, typical controllers execute control algorithms at high speeds to quickly correct undesirable deviations in the process. Undesirable environmental factors or other adverse conditions may create intermittent interference that disrupts or prevents the high speed communications necessary to support such execution of monitoring and control algorithms. Fortunately, wireless networks have become much more robust over the past two decades, enabling the reliable use of wireless communications in some types of process control systems.

However, power consumption remains a complicating factor when using wireless communication in process control applications. Because wireless field devices are physically separated from the I/O network, the field devices typically need to provide their own power source. Thus, field devices can be battery powered, draw solar energy, or harvest environmental energy such as vibration, heat, pressure, etc. For these devices, the energy consumed for data transmission can be a significant portion of the total energy consumption. In fact, more power may be consumed during the process of establishing and maintaining a wireless communication connection than during other critical operations performed by the field device, such as steps taken to sense or detect the process variable being measured. To reduce the power consumption of wireless process control systems and thus extend battery life, it has been proposed to implement wireless process control systems in which field devices, such as sensors, communicate with a controller in a non-periodic, slow, or intermittent manner. In one instance, the field device may communicate with or transmit measurements of a process variable to a controller only when a significant change in the process variable is detected, leading to a non-periodic communication with the controller.

One control technique developed to handle non-periodic process variable measurement updates uses a control system that provides and maintains an indication of the expected process response to control signals generated by the controller during infrequent or non-periodic measurement updates. The expected process response can be developed by a mathematical model that calculates the expected process response to the control signal for a given measurement update. An example of this technique is described in U.S. Pat. No. 7,587,252, entitled "Non-Periodic Control Communications in Control Systems," the entire disclosure of which is expressly incorporated herein by reference.
A control system that generates an indication of an expected process response to a control signal upon receiving a non-periodic process variable measurement update and has a filter that maintains the generated indication of the expected process response until the arrival of the next non-periodic process variable measurement update. As another example, U.S. Pat. No. 7,620,460, entitled "Process Control With Unreliable Communications," the disclosure of which is expressly incorporated herein by reference in its entirety, discloses a system that provides an indication of an expected response to a control signal but includes a filter that further modifies the filter to incorporate a measurement of the time that has elapsed since the last non-periodic measurement update to generate a more accurate indication of the expected process response.

However, in many control applications, the process control system may receive set point changes during process operation. Generally, during the execution of a periodically updated control system (e.g., a hardwired control communication system), when the set point changes, a controller designed to take proportional action to the error between the set point and the measured process variable will immediately change the controller output to drive the process variable towards a new steady state value. However, in a wireless control system receiving infrequent, non-periodic measurement updates operating as described in both examples above, the measured process response reflected by each new measurement update will reflect the change made in the controller output captured for the last measurement update in addition to the change in the output caused by a set point change made some time after the last measurement update was received. In this case, calculating the reset component of the controller based on the controller output and the time since the last measurement update (as described in U.S. Pat. No. 7,620,460) may overcompensate for the changes made since the last measurement update. Thus, the process response to a set point change may differ based on when the set point change was made after the last measurement update. As a result, this system does not respond quickly or robustly to setpoint changes because when the controller generates a control signal after a setpoint change, it continues to rely on a previously generated (and now outdated) indication of the expected response. To solve this problem, U.S. Patent Application Publication No. 2013/0184837, entitled "Compensating for Setpoint Changes in a Non-Periodically Updated Controller," the entire disclosure of which is expressly incorporated by reference, discloses a system that uses a continuously updated filter in a feedback loop within a controller to track the behavior of a controlled variable during times when no new process variable measurements are received, and uses the output of the controller when a new process variable measurement is received at the controller, but otherwise uses the output of the filter from the most recent time that a process variable measurement was received to generate a control signal. In this system, the generated control signal responds better and performs better when the set point of the process variable changes between the times that the process variable measurements are received.

Furthermore, when using battery-powered transmitters in a wireless control system, it is desirable to set up the system to maintain long battery life. For example, to obtain a 3-5 year battery life using current transmitter and battery technology, it is generally necessary to use a communication update rate of 8 seconds or faster. However, using such slow update rates limits the use of PID (Proportional-Integral-Derivative) based wireless control for processes with process response times of 30 seconds or longer, since in order to maintain control of the process, it is still important to have process feedback measurements received at a rate that is at least 4 times the rate associated with the process response time (i.e., the inverse of the process response time).

Still further, there are various types of PID algorithms that can be used to address wireless control, including those that use to generate a position output, e.g., a 4-20 ma signal or a digital signal, that is provided to a valve or other controlled element that indicates the final position to which it should move. However, there are also other PID algorithms that provide a velocity-based control signal that instructs a valve or other controlled element to move a certain amount in a certain direction, e.g., by energizing a movable element for a certain amount of time. Such velocity-based control signals are commonly used with electric motors and provide the control signal in the form of a pulse signal (with the pulse width modulated to indicate the amount of time the valve must be energized to move the movable element for a certain period of time). Velocity-based controllers tend to generate a change in position signal rather than a signal that is indicative of the actual position attained by the movable element. Thus, velocity-based control algorithms tend to be used to provide incremental (increase/decrease) outputs to actuators, and thus can be used in controlling actuators that cannot provide position feedback.

The control technique allows for robust control of a process or control loop that has fast dynamics with respect to the rate at which measurements of the controlled process variable are provided as feedback to a process controller. Specifically, the control technique can control a process using a PID algorithm in the form of a position or velocity, with measurements or feedback signals of the process variable provided to the controller at time intervals equal to or even longer than the process response time. Specifically, the control technique can be used to provide robust control in processes that have response times 2-4 times shorter than the feedback time interval. Such situations can arise, for example, when using wireless control where feedback measurements of the process variable are provided to the controller wirelessly, intermittently, or at time intervals shorter than, close to, or longer than the response time of the process.

The disclosed velocity PID control routine can be used in many different situations, such as for control using wireless measurements when an actuator requires incremental input and no position feedback is available to the controller, for control by wired measurements when interfacing to actuators that require position or incremental input, and for addressing legacy installations as well as new installations where wireless measurements are used for control. Furthermore, adjustments of the velocity form PID algorithm can be made based on process gains and dynamics and are independent of wireless communication rates. Still further, the velocity form PID control routine automatically retains the last output position upon loss of communication and provides a bumpless recovery when communication is re-established.

In one case, a controller implementing the new control technique generally includes a different structure that generates differential, proportional, integral, and derivative control signal components and uses them to generate a differential or displacement-based control signal that is then sent to a controlled device, such as a valve, to control the operation of the controlled device and therefore the process. This differential or velocity-based form of control generates a control signal that performs better than standard PID control in the presence of feedback measurements of slow process variables. Specifically, a controller using this control technique generates a differential proportional value during each controller iteration that represents the difference between the previous proportional control signal component and the newly calculated proportional control signal component, and this differential proportional value is used as the basis for each new control signal from the controller. However, when new process variable measurement signals are available to the controller, various other control signal components, such as a differential control signal component and/or an integral control signal component, can be added to or combined with the differential proportional control signal component. These two control signal components can also be based on the difference between the newly calculated value and the previously calculated value. Specifically, a new differential component may be calculated during a controller iteration, at which point a newly received value of the process variable measurement is made available to the controller. Similarly, a new integral component may be generated using a continuously updated filter, which generates a new indication of the expected response of the process to each control routine iteration of the controller. However, the output of the continuously updated filter is used to generate a new integral element only when a new value of the process variable measurement is received. At all other times, the integral control signal component is set to zero.

The rate-based PID control techniques disclosed herein use a differential form to generate a control signal that rapidly adapts to set point changes (even during the time that the process variable feedback signal is present at the controller input), yet still provide robust and stable control in the presence of slowly received (e.g., intermittent) feedback signals, including feedback signals received at a rate slower than (e.g., 2-4 times slower than), close to, or faster than the inverse of the response time of the process being controlled.

FIG. 1 is a block diagram of an exemplary periodically updated hardwired process control system. 1 is a graph illustrating process output response to process input of an exemplary periodically updated hardwired process control system, including process response time; FIG. 1 is a block diagram illustrating an example of a wireless process control system having a controller that receives slow or non-periodic feedback input. FIG. 1 is a block diagram of an example controller that enables robust compensation of set point changes or feedforward disturbances in a periodically updated hardwired process control system. 4B is a graph illustrating the process output response of the example controller of FIG. 4A as the controller responds to several set point changes. FIG. 1 is a block diagram of an example controller that provides compensation for set point changes in a non-periodically updated process control system, where the controller compensates for process and/or measurement lags in the feedback signal. FIG. 1 is a block diagram of an example controller that provides compensation for set point changes in a non-periodically updated process control system in which the process controller determines a control signal using a derivative contribution or a rate contribution. FIG. 1 is a block diagram of an example controller that compensates for set point changes in a non-periodically updated process control system in which a process controller receives additional controller input data provided from field devices, control elements, or other downstream devices to affect the operational response of the process. FIG. 1 is a block diagram of an example controller that provides compensation for set point changes in a non-periodically updated process control system in which a process controller adapts to use either actual or implied controller input data for field devices. FIG. 1 is a block diagram of an example speed-based PI controller that enables robust compensation of set point changes or feedforward disturbances in a process control system in response to a process variable measurement signal received at a slow speed. FIG. 1 is a block diagram of an example speed-based PID controller that enables robust compensation of set point changes or feedforward disturbances in a process control system in response to a process variable measurement signal received at a slow speed. 1 is a graph illustrating the simulated process response of an exemplary prior art speed-based PID controller in both wired and wireless configurations in response to a set point change in a primary controlled variable and having a process response time of 8 seconds. 11 is a graph illustrating a simulated process response of an exemplary inventive speed-based PID controller in both wired and wireless configurations in response to a set point change in a primary control variable and having a process response time of 8 seconds. 1 is a graph illustrating simulated process response of an exemplary prior art speed-based PID controller in both wired and wireless configurations in response to a set point change in a primary controlled variable and having a process response time of 3 seconds. 1 is a graph illustrating a simulated process response of an exemplary inventive speed-based PID controller in both wired and wireless configurations in response to a set point change in a primary control variable and having a process response time of 3 seconds; 1 is a graph illustrating a simulated process response of an exemplary prior art speed-based PID controller in both wired and wireless configurations in response to a disturbance change and having a process response time of 8 seconds; 11 is a graph illustrating a simulated process response of an exemplary velocity-based PID controller according to the present invention in both wired and wireless configurations in response to a disturbance change and having a process response time of 8 seconds. 1 is a graph illustrating a simulated process response of an exemplary prior art speed-based PID controller in both wired and wireless configurations in response to a disturbance change and having a process response time of 3 seconds; 11 is a graph illustrating a simulated process response of an exemplary velocity-based PID controller according to the present invention in both wired and wireless configurations in response to a disturbance change and having a process response time of 3 seconds.

The control technique can be used to provide control in process loops in which process measurement feedback signals are received slowly or intermittently, and is particularly useful when the process measurement feedback signals are received at a rate slower than, comparable to, or slightly faster than a rate associated with the process dynamics being controlled, such as the inverse of the process response time. The controller can be used, for example, in controllers that receive process measurement signals as feedback signals in a slow and/or non-periodic manner, particularly at a rate that is less than, comparable to, or otherwise similar to the process response rate of the process dynamics being controlled (i.e., the inverse of the process response time to the process variable being controlled). In one case, the control technique generates a control signal for use in controlling a process device, such as a valve, by combining a proportional contribution signal with one or more of a differential contribution signal and an integral contribution signal. In one case in which a speed-based PID algorithm is used, the controller generates a proportional contribution value from the difference between the set point and the most recently received feedback signal of the process variable measurement. This error signal is then provided to a differential unit which multiplies it by a gain signal to determine the change in this signal since the last execution cycle of the controller. Additionally, a differential unit may receive the error signal and essentially differentiate the error signal over time since the last measurement signal was received at the controller to perform a differential calculation on the error signal. The output of the differential module or calculation is also provided to a change detection unit or differential unit to determine the difference between the current output of the differential calculation and the previous value used to calculate the control signal. Similarly, an integral calculation unit receives and integrates (e.g., sums) the difference in the control signal as a product of the proportional and differential units. The output of the summer is provided to a filter to filter the output of the summer to generate an integral contribution signal. However, this integral contribution is provided to the summer which sums it with the output of the control signal only when a new feedback value is provided to the controller. That is, the integral contribution is set to zero for all controller iterations except those where a new feedback signal is available at the controller.

Generally speaking, the continuously updated filter of the integral contribution unit in the controller generates an indication of the expected process response (also called feedback contribution) during each control routine iteration of the controller, despite the slow or non-periodic receipt of process variable measurement updates from the field device. The continuously updated filter uses, in part, a previously generated indication of the expected response from the last control routine iteration and control routine execution period to generate an indication of the expected response during each control routine iteration. In addition, the integral output switch in the controller provides the output of the continuously updated filter as a feedback contribution, such as an integral (also known as reset) contribution based on the latest measurement indication, to the control signal. Generally speaking, the integral output switch provides the expected process response, as generated by the continuously updated filter at the time the last measurement update was received by the controller in the form of a speed PID controller, or a zero value, as an integral or reset contribution to the control signal during each iteration of the controller where no new measurement signal is available. When a new measurement update is available, the integral output switch clamps to the new indication of the expected process response generated by the continuously updated filter (based on the indication of the new measurement update) and provides the new expected process response as the integral contribution of the control signal. As a result, the controller uses the continuously updated filter to determine a new expected response of the process during each controller iteration, and each new expected process response reflects the effects of set point or feedforward changes that have been made in the time between measurement updates and that affect the controller output during the evolution of the control signal, even though the integral or reset component of the control signal changes only when a new measurement update is available to the controller.

1 illustrates a process control system 10 that can be used to implement the described control method. The process control system 10 includes a process controller 11 that can be connected to a data historian 12 and to one or more host workstations or computers 13 (which can be any type of personal computer, workstation, etc.), each having a display screen 14. The controller 11 is also connected to field devices 15-22 via input/output (I/O) cards 26 and 28. The data historian 12 can be any desired type of data collection unit having any desired type of memory for storing data, and any desired or known software, hardware, or firmware. In FIG. 1, the controller 11 is communicatively connected to the field devices 15-22 using a hardwired communication network and communication scheme.

Generally, the field devices 15-22 can be any type of device, such as sensors, valves, transmitters, positioners, etc., while the I/O cards 26 and 28 can be any type of I/O device conforming to any desired communication or controller protocol. The controller 11 includes a processor 23 that implements or oversees one or more process control routines (or any modules, blocks, or subroutines thereof) stored in memory 24. Generally speaking, the controller 11 communicates with the devices 15-22, the host computer 13, and the data historian 12 to control the process in any desired manner. Furthermore, the controller 11 implements control strategies or schemes using what are generally referred to as function blocks, each of which is an object or other portion (e.g., a subroutine) of an overall control routine that operates with other function blocks (through communications called links) to implement a process control loop within the process control system 10. A function block typically performs one of the following functions to perform some physical function within the process control system 10: an input function, such as associated with a transmitter, sensor, or other process parameter measurement device, a control function, such as associated with a control routine performing PID, fuzzy logic control, etc., or an output function to control the operation of some device, such as a valve. Of course, hybrid and other types of function blocks exist and can be utilized herein. Function blocks can be stored in and executed by the controller 11 or other devices, as described below.

As illustrated in decomposition block 30 of FIG. 1, controller 11 may include several single-loop control routines, illustrated as control routines 32 and 34, and may also implement one or more advanced control loops, illustrated as control loop 36, if desired. Each control loop is typically referred to as a control module. Single-loop control routines 32 and 34 are illustrated to provide single-loop control using single-input/single-output fuzzy logic control blocks and single-input/single-output PID control blocks, respectively, connected to appropriate analog input (AI) and analog output (AO) function blocks that may be associated with process control devices such as valves, measurement devices such as temperature and pressure transmitters, or any other devices within process control system 10. Although the advanced control loop 36 is illustrated as including an advanced control block 38 having inputs communicatively connected to one or more AI function blocks and outputs communicatively connected to one or more AO function blocks, the inputs and outputs of the advanced control block 38 may be connected to any other desired function blocks or control elements to receive other types of inputs and provide other types of control outputs. The advanced control block 38 may implement any type of multiple-input, multiple-output control scheme and may constitute or include a model predictive control (MPC) block, a neural network modeling or control block, a multivariable fuzzy logic control block, a real-time optimizer block, etc. It will be understood that the function blocks illustrated in FIG. 1, including the advanced control block 38, may be executed by a stand-alone controller 11 or, alternatively, may be located and executed within any other processing device or control element of a process control system, such as one of the workstations 13 or one of the field devices 19-22. As an example, field devices 21 and 22 may be a transmitter and a valve, respectively, and may execute control elements for implementing control routines, and thus may include processing components and other components for implementing portions of the control routines, such as one or more function blocks. More specifically, as illustrated in FIG. 1, field device 21 may have a memory 39A for storing logic and data associated with an analog input block, while field device 22 may include an actuator having a memory 39B for storing logic and data associated with a PID or other control block in communication with an analog output (AO) block.

The graph of Figure 2 generally illustrates process outputs that are developed in response to process inputs of a process control system based on the implementation of one or more of control loops 32, 34, and 36 (and/or any control loops incorporating function blocks present in field devices 21 and 22 or other devices). The implemented control routines generally execute in a periodic manner through a number of controller iterations with the number of control routine executions shown along the time axis by thick arrow 40 in Figure 2. In the conventional case, each control routine iteration 40 is supported by updated process measurements shown by thin arrow 42, for example provided by a transmitter or other field device. As illustrated in Figure 2, there are typically multiple periodic process measurements 42 made and received by the control routine during each periodic control routine execution time 40. To avoid limitations associated with synchronizing measurements and control execution, many known process control systems (or control loops) are designed to oversample process variable measurements by a factor of 2 to 10 times. Such oversampling helps ensure that process variable measurements for use in the control scheme are current during each control routine execution or iteration. Also, to minimize control drift, conventional designs dictate that feedback-based control should execute 4-10 times faster than the process response time, and that new process variable measurements are available at each controller iteration. The process response time is represented in the process output response curve 43 of the graph in FIG. 2 as being the time associated with the process time constant (τ) (e.g., 63% of the change in the process variable) plus the process delay or dead time (T _D ) (shown by the lower line 45 in FIG. 2 ) following the implementation of a step change 44 in the process input. In any event, to meet these conventional design requirements, the process measurement updates (shown by arrow 42 in FIG. 2 ) are sampled and provided to the controller at a rate that is much faster than the control routine execution rate (shown by arrow 40 in FIG. 2 ), and thus much faster or higher than the process response time.

However, obtaining frequent periodic measurement samples from a process may not be practical or even possible, for example, when a controller is operating in a process control environment where the controller wirelessly receives measurements from one or more field devices. Specifically, in these cases, the controller may only be able to receive slow or non-periodic process variable measurements (to conserve battery life of the wireless sensor/transmitter). Furthermore, in these cases, the time between non-periodic or even periodic process variable measurements may be longer than the control routine execution rate (indicated by arrow 40 in FIG. 2). FIG. 3 depicts an exemplary wireless process control system 10 capable of implementing slow and/or non-periodic wireless communication of process control data or use of process variable measurements in the controller 11.

3 is essentially similar to the control system 10 of FIG. 1, with like elements numbered the same. However, the control system 10 of FIG. 3 includes a number of field devices 60-64 and 71 wirelessly communicatively coupled to the controller 11 and potentially to each other. As illustrated in FIG. 3, the wirelessly connected field device 60 is connected to an antenna 65 and cooperates to wirelessly communicate with an antenna 74, which is in turn coupled to a wireless I/O device 68. Further, the field devices 61-64 are connected to a wired-to-wireless conversion unit 66, which is in turn connected to an antenna 67. The field devices 61-64 communicate wirelessly through the antenna 67 with an antenna 73, which is connected to a further wireless I/O device 70. Also illustrated in FIG. 3, the field device 71 includes an antenna 72 that communicates with one or both of the antennas 73 and 74, thereby communicating with the I/O devices 68 and/or 70. I/O devices 68 and 70 are then communicatively connected to controller 11 via a wired backplane connection (not shown in FIG. 3). In this case, field devices 15-22 remain hardwired to controller 11 via I/O devices 26 and 28.

The process control system 10 of FIG. 3 generally uses wireless transmission of data measured, sensed, or calculated by the transmitters 60-64 or other control elements such as the field device 71, as described below. In the control system 10 of FIG. 3, new process variable measurements or other signal values are considered to be transmitted by the devices 60-64 and 71 to the controller 11 on a slow or non-periodic basis, such as when certain conditions are met. For example, new process variable measurements can be transmitted to the controller 11 when the process variable value has changed by a predetermined amount relative to the last process variable measurement transmitted to the controller 11 by the device. These signals can also be transmitted periodically, but typically at a much slower rate for typical process control systems such as wired process control signals. For example, the slow periodic feedback rate can be lower than the controller execution rate (the rate at which the controller generates new control signals for use in creating the control signals) and can be at a rate lower than, on the same level as, or similar to the process response rate or response time, such as a rate 2-4 times lower than the response rate of the process dynamics being controlled using the control techniques described herein. Here, the process response rate is the inverse of the process response time. Of course, other methods of determining when to transmit process variable measurements in a periodic or aperiodic manner may also or alternatively be implemented.

As will be appreciated, the transmitters 60-64 of FIG. 3 can transmit signals indicative of the respective process variables (e.g., flow, pressure, temperature, or level signals) to the controller 11 for use in one or more control loops or control routines, or for use in monitoring routines. Other wireless devices, such as the field device 71, can wirelessly receive process control signals and/or can be configured to transmit other signals indicative of any other process parameters. Generally speaking, as illustrated in FIG. 3, the controller 11 includes a communications stack 80 running on a processor that processes incoming signals, a module or routine 82 running on the processor to detect when the incoming signal includes a measurement update, and one or more control modules 84 running on the processor to perform control based on the measurement update. The detection routine 82 can generate a flag or other signal to indicate that the data provided via the communications stack 80 includes a new process variable measurement or other type of update. The new data and update flags can then be provided to one or more of the control modules 84 (which can be function blocks), which are then executed by the controller 11 at a predetermined periodic execution rate, as will be described in more detail below. Alternatively, or in addition, the new data and update flags can be provided to one or more monitoring modules or applications executing in the controller 11 or in other parts of the control system 10.

3 generally results in slow or periodic, irregular or otherwise infrequent data transmission between the field devices 60-64 and 71 and the controller 11. However, as noted above, communication of measurements from the field devices 15-22 to the controller 11 is traditionally structured to occur in a periodic fashion at a rate much faster than the controller's execution rate, or at least much faster than the dynamic rate of the process, i.e., the inverse of the process response time (for the process phenomenon being controlled). As a result, the control routines of the controller 11 are generally designed for periodic updates of process variable measurements used in the feedback loops of the controller 11.

To accommodate slow, non-periodic, or otherwise unavailable measurement updates (and other unavailable communication transmissions), such as those introduced by wireless communication between some of the field devices and the controller 11, the control and monitoring routine(s) of the controller 11 can be restructured or modified to enable the process control system 10 to function properly when using slow, including non-periodic, or intermittent updates, as described below, particularly when these updates occur less frequently than the execution rate of the controller 11, and further when receiving such updates at a rate similar (e.g., 2-4 times slower or the same rate, etc.) to the process response rate (e.g., the inverse of the process response time of the process variable being controlled). FIGS. 4-10 illustrate in further detail an exemplary control scheme configured to operate using slow and/or non-periodic control-related communications. For example, FIG. 4A generally illustrates a position-type process controller 100 coupled to a process 101. The control scheme implemented by the controller 100 (which may be the controller 11 of FIGS. 1 and 3 or a control element of a field device, such as one of the wireless field devices of FIG. 3) generally includes the functionality of the communications stack 80 illustrated and described in connection with FIG. 3, an update detection module 82, and one or more of a control module 84, and generates control signals indicative of the position to which a movable element of the control device should move.

In the exemplary system of FIG. 4A, the controller 100 operates to receive a setpoint signal, for example from one of the workstations 13 (FIGS. 1 and 3), or from any other source within or in communication with the process control system 10, and generate one or more control signals 105 that are provided to the process 101 from an output of the controller 100. In addition to receiving the control signal 105, the process 101 may be subject to measured or unmeasured disturbances, as shown diagrammatically by arrows 104. Depending on the type of process control application, the setpoint signal may be changed at any time during the control of the process 101, such as by a user, a tuning routine, or the like. Of course, the process control signal 105 may control an actuator associated with a valve, or any other field device to affect the response of the operation of the process 101. The response of the process 101 to the change in the process control signal 105 is measured or sensed by a transmitter, sensor, or other field device 106, which may correspond, for example, to any one of the transmitters 60-64 illustrated in FIG. 3. The communication link between the transmitter 106 and the controller 100 may include a wireless connection and is illustrated in FIG. 4A using dashed lines.

In a simple embodiment, the controller 100 can implement a single-input, single-output closed-loop control routine, such as a PI control routine, which is a form of a PID control routine. Thus, the controller 100 includes several standard PI controller elements, including a communication stack 80, and a control signal generator, including a summing block 108, a proportional gain element 110, a further summing block 112, and a high and low limiter 114. The control routine 100 also includes a direct feedback path, including a filter 116 and an integral output switch, including a selection block 118. The filter 116 is coupled to the output of the high and low limiter 114, and the block of switches 118 is coupled to the output of the filter 116 to provide the summing block 112 with an integral or reset contribution or component of the control signal being generated by the controller 100.

During operation of the controller 100, summing block 108 compares the set point signal with the most recently received process variable measurement provided from the communications stack 80 within the controller 100 to generate an error signal. Proportional gain element 110 operates on the error signal, for example by multiplying the error signal by a proportional gain value _Kp , to generate a proportional contribution or component of the control signal. Summing block 112 then combines the output of gain element 110 (i.e., the proportional contribution) with the integral or reset contribution, or component of the control signal generated by the feedback path, to generate an essentially unlimited control signal. Limiter block 114 then performs high and low limiting on the output of summing block 112 to generate the control signal 105 that is sent to control the process 101.

Importantly, the filter 116 and the block or switch 118 in the feedback path of the controller 100 operate in the following manner to generate an integral or reset contribution to the control signal. The filter 116, coupled to receive the output of the limiter 114, generates an indication of the expected process response to the control signal 105 based on the output value of the limiter 114 and the duration or time of execution of the control algorithm 100. The filter 116 provides this expected process response signal to the switch or block 118. The switch or block 118 samples and clamps the output of the filter 116 at the output of the switch or block 118 whenever a new process variable measurement is received, and maintains this value until the next process variable output is received at the communication stack 80. Thus, the output of the switch 118 remains the output of the filter 116 sampled at the last measurement update.

The expected process response to a change in the output of summer 108, as produced by filter 116, can be approximated using a first-order model, as described in more detail below. More generally, however, the expected process response can be generated using any suitable model of process 101 and is not limited to a model incorporated into the feedback path of controller 100 or to a filter or model associated with determining an integral or reset contribution to the control signal. For example, a controller utilizing a model to provide an expected process response can incorporate a derivative contribution such that control routine 100 implements a PID control scheme. Some examples incorporating exemplary types of derivative contributions are described below in connection with FIGS. 6-8.

Before discussing the operation of filter 116 of FIG. 4A in more detail, it is useful to note that a conventional PI controller can be implemented using a positive feedback network to determine the integral or reset contribution. Mathematically, it can be seen that the transfer function for a conventional PI implementation is equivalent to the standard equation for unconstrained control, i.e., when the output is not limited. In particular:



One advantage of using a positive feedback path in the controller 100, as shown in FIG. 4A, is that the controller output is automatically prevented from winding up the reset contribution when it is high or low, i.e., limited by limiter 114.

In any event, the control technique described below allows the controller to use a positive feedback path to determine the reset or integral contribution when it receives a non-periodic update of the process variable, yet still allows for a robust controller response in the event of set point or feedforward changes occurring between receipt of new process variable measurements. Specifically, to provide robust set point changes to the controller operation, the filter 116 is configured to calculate a new indication or value of the expected process response during each or every execution of the controller 100, regardless of whether this output of the filter is provided to the summing block 112. As a result, the output of the filter 116 is newly regenerated during each execution cycle of the controller routine, even if only the output of the filter 116 generated immediately after the controller 100 receives a new process measurement update from the communication stack 80 is used as the integral or reset contribution in the summing block 112.

Specifically, a new indication of expected response as produced by filter 116 is calculated during each controller execution cycle from the current controller output (i.e., the control signal after limiter 114), the indication of expected response produced by filter 116 during the last (i.e., immediately preceding) controller execution cycle, and the duration of the controller execution. As a result, filter 116 is described herein as being continuously updated as it executes to generate a new estimate of the process response during each controller execution cycle. Exemplary equations that may be implemented by the continuously updated filter 116 to generate a new expected process response or filter during each controller execution cycle are set forth below.

It will now be noted that the new filter output F _N is iteratively determined by adding a damping component determined by adding the difference between the current controller output value O _{N-1 and} the current filter output value F _N-1 multiplied by a coefficient that depends on the reset time T _Reset and the execution period ΔT of the controller. By using a filter that is continuously updated in this manner, the control routine 100 is able to better determine the expected process response when it calculates the integral control signal input upon receiving a new process variable measurement, thereby being more sensitive to changes in the set point or other feedforward disturbances that occur between the receipt of two process variable measurements. More specifically, it will be noted that a change in the set point (without the receipt of a new process measurement) immediately results in a change in the error signal at the output of the summer 108, which changes the proportional contribution of the control signal and therefore changes the control signal. As a result, the filter 116 immediately begins to generate a new expected response of the process to the changed control signal, and therefore updates its output, before the controller 100 receives a new process measurement. Then, when the controller 100 receives a new process measurement and a sample of the filter output is clamped by switch 118 to the input of summer 112 for use as an integral or reset contribution to the control signal, the filter 116 is repeating, at least to some extent, the expected process response that reacts to or incorporates the response of the process 101 to the set point change.

In the past, reset contribution filters used in the feedback path of non-periodically updated controllers, such as in the systems described in U.S. Patents 7,587,252 and 7,620,460, only calculated a new indication of expected response when a new process variable measurement was available. As a result, the reset contribution filter did not compensate for setpoint changes or feedforward disturbances that occurred between receipt of a process variable measurement, because the setpoint changes or feedforward disturbances were completely independent of any measurement updates. For example, if a setpoint change or feedforward disturbance occurred between two measurement updates, the expected process response of the controller could be distorted because the calculation of a new indication of expected response was based on the time since the last measurement update and the current controller output 105. As a result, the filter 116 could not begin to compensate for time changes in the process (or control signal) that resulted from a setpoint change (or other feedforward disturbance) that occurred between receipt of two process variable measurements at the controller.

However, as will be appreciated, the control routine 100 of FIG. 4A provides an expected process response by basing the calculation of non-periodic measurements on the expected response between receipt of two measurements to compensate for changes caused by set point changes (or any measured disturbances used as feedforward inputs to the controller 100). Thus, the control technique described above can accommodate set point changes, feedforward actions to measured disturbances, etc. that may affect the expected process response, thus providing a more robust control response.

As can be seen, the control technique illustrated in FIG. 4A calculates an indication of the expected response via a continuously updated filter 116 (e.g., a reset contribution filter) for each execution of the control block or routine 100. Here, the controller 100 configures the continuously updated filter 116 to calculate a new indication of the expected response for each execution of the control block. However, to determine whether the output of the filter 116 should be used as an input to the summing block 112, the communication stack 80 and, in some embodiments, the update detection module 82 (FIG. 3) process the incoming data from the transmitter 106 to generate a new value flag for the integral output switch 118 when a new process variable measurement is received. This new value flag informs the switch 118 to sample and clamp the filter output value for this controller iteration to the input of the summer 112.

Whether or not a new value flag is communicated, the continuously updated filter 116 continues to calculate an indication of the expected response for each iteration of the control routine. This new indication of the expected response is conveyed to the integral output switch 118 for each execution of the control block. Depending on the presence of the new value flag, the integral output switch 118 switches between allowing the new indication of the expected response from the continuously updated filter 116 to pass to the summing block 112 and maintaining the signal previously delivered to the summing block 112 during the last execution of the control block. More specifically, when a new value flag is communicated, the integral output switch 118 allows the last calculated indication of the expected response from the continuously updated filter 116 to pass to the summing block 112. Conversely, if the new value flag is not present, the integral output switch 118 retransmits the indication of the expected response from the last control block iteration to the summing block 112. In other words, the integral output switch 118 clamps to a new indication of the expected response each time a new value flag is communicated from the stack 80, but does not allow any newly calculated indication of the expected response to reach the summing block 112 if the new value flag is not present.

This control technique allows the continuously updated filter 116 to continue to model the expected process response regardless of whether new measurements are communicated. If the control output changes as a result of a set point change or a feedforward action based on a measured disturbance, the continuously updated filter 116 correctly reflects the expected process response by calculating a new indication of the expected response at each control routine iteration, regardless of the presence of the new value flag. However, the new indication of the expected response (i.e., the reset contribution or integral component) is only incorporated into the controller calculations when the new value flag is communicated (via the integral output switch 118).

The graph 200 illustrated in Figure 4B depicts the simulated operation of the controller 100 of Figure 4A in driving the process output signal 202 to a steady state value as the controller 100 responds to several set point changes. In Figure 4B, the process output signal 202 (illustrated by a thick line) is shown compared to the set point signal 204 (illustrated by a thin line) during wireless operation of the process control system. As indicated by the arrows along the time axis at the bottom of the graph 200, when a set point change occurs, the controller 102 responds to react to the new set point (i.e., the steady state value) by generating a control signal that drives the process output. For example, as illustrated in Figure 4B, a set point change occurs at time _T1 , which is evident from a significant change in the magnitude of the set point signal 204 from a higher value to a lower value. In response, the controller 102 drives the process variable associated with the set point or set point value for the new steady state in a smooth transient curve as shown by the output signal 202 between times _T1 and _T2 . Similarly, in FIG. 4B, a second set point change occurs at time _T2 , which is evident from the large change in magnitude of the set point signal 204 from a lower value to a higher value. In response, the controller 102 controls the process variable associated with the set point or set point value for the new steady state in a smooth transient curve as shown by the output signal 202 between times _T2 and _T3 . As a result, as can be seen in FIG. 4B, the controller 100 implementing the control routine described above enables compensation of set point changes in a non-periodic wireless control system in a robust manner. Because feedforward disturbances can be measured and included in the control action operation, the controller 100 implementing the control routine described above can also enable compensation of feedforward changes in the control output in a non-periodic wireless control system.

Note that the simple PI controller configuration of FIG. 4A directly uses the output of the filter 116 as the reset contribution to the control signal, and in this case the reset contribution of the closed loop control routine (e.g., the continuously updated filter equation presented above) can provide an accurate representation of the process response in determining whether the process exhibits steady state operation. However, other processes, such as processes dominated by dead time, may require incorporating additional components into the controller of FIG. 4A to model the expected process response. For processes that are adequately represented by a first-order model, the process time constant can generally be used to determine the reset time of the PI (or PID) controller. More specifically, if the reset time is set equal to the process time constant, the reset contribution generally overrides the proportional contribution so that the control routine 100 reflects the expected process response over time. In the example illustrated in FIG. 4A, the reset contribution can be achieved by a positive feedback network with a filter having a time constant the same as the process time constant. Although other models can be used, the positive feedback network, filter, or model provides a convenient mechanism for determining the expected response of a process with a known or approximated process time constant. For processes requiring PID control, the derivative contribution, also known as rate, to the PID output may be recalculated and updated only when a new measurement is received. In such cases, the derivative calculation may use the elapsed time since the last new measurement. Although other controller components can be used to control more complex processes using non-periodic receipt of process measurements, several examples of controllers that can use the filtering technique of FIG. 4A to provide robust control in response to set point changes are described below in conjunction with FIGS. 5-8.

5, an alternative controller (or control element) 120, configured according to the control technique as described above, is similar in many respects to the controller 100 illustrated in FIG. 4A. As a result, elements common to both controllers are identified with the same reference numbers. However, the controller 120 incorporates additional elements into a control routine that determines the expected process response during the transmission of measurements. In this case, the process 101 can be characterized as having a significant amount of dead time, and as a result, a dead time unit or block 122 is included in the controller model for dead time correction. The incorporation of the dead time unit 122 generally helps to arrive at a more accurate representation of the process response. More specifically, the dead time unit 122 can be implemented in any desired manner and can include or utilize methods common to Smith predictors or other known control routines. However, in this situation, the continuously updated filter 116 and switch module 118 operate in the same manner as described above with respect to the controller 100 of FIG. 4A to provide robust control in response to set point changes.

FIG. 6 depicts another alternative controller (or control element) 130, which differs from the controller 100 described above in FIG. 4A in that a derivative or rate contribution component is incorporated into the controller 130. By incorporating a derivative contribution, the control routine implemented by the controller 130 includes an additional feedback mechanism, such that in some cases a proportional-integral-derivative (PID) control scheme is implemented.

The control routine or technique of FIG. 6 includes a derivative contribution configured in a manner similar to that described above in connection with the integral contribution of FIG. 4A to accommodate non-periodic or otherwise unavailable updates of the process measurement. The derivative contribution can be reconfigured based on the time elapsed since the last measurement update. In this manner, spikes in the derivative contribution (and the resulting output signal) are avoided. More specifically, the derivative contribution of FIG. 6 is determined by a derivative block 132, which receives an error signal from summing block 108 in parallel with elements dedicated to the proportional and integral contributions. The proportional, integral, and derivative contributions are combined in summing block 134 as shown in FIG. 6, although other PID configurations (e.g., serial configurations) can also be utilized.

To accommodate unreliable transmissions and, more generally, unavailability of measurement updates, the derivative contribution is maintained at the last determined value until a measurement update is received, as indicated by a new value flag from the communications stack 80. This technique allows the control routine to continue periodic execution according to the control routine's normal or established execution rate. Upon receiving an updated measurement, as illustrated in FIG. 6, the derivative block 132 may determine the derivative contribution according to the following equation:

This technique for determining the derivative contribution allows measurement updates to the process variable (i.e., the control input) to be lost for one or more execution periods without producing output spikes. When communication is re-established, the term (e _N -e _N-1 ) in the derivative contribution equation can produce the same value as it would in a standard calculation of the derivative contribution. However, for standard PID techniques, the divisor in determining the derivative contribution is the execution period. In contrast, the present control technique utilizes the elapsed time between two successfully received measurements. By using an elapsed time that is longer than the execution period, the control technique produces a smaller derivative contribution than the standard PID technique and also reduces spiking.

To facilitate the determination of the elapsed time, the communication stack 80 can provide the new value flag described above to the derivative block 132 as shown in FIG. 6. Alternative embodiments can include or involve detection of new measurements or updates based on the value. Also, the process measurement can be used in place of the error in the calculation of the proportional or derivative components. More generally, the communication stack 80 can include or involve any software, hardware, or firmware (or any combination thereof) to implement a communication interface with the process 101, including any field devices within the process 101, process control elements external to the controller, etc. However, in the controller 130 of FIG. 6, the continuously updated filter 116 and switch module 118 operate in the same manner as described above with respect to the controller 100 of FIG. 4A to provide robust control in response to set point changes.

Actuators or other downstream elements controlled by the controllers described in connection with Figures 3, 4A, and 5-6 may nevertheless receive control signals with abrupt changes, particularly after a period of time without any communication between the controller or control element and the downstream actuators or other elements. The resulting control action may in some cases be abrupt enough to affect plant operation, and such abrupt changes may lead to inappropriate levels of instability.

The possibility of sudden control changes due to loss of communication between the controller and downstream elements can be addressed by incorporating actual downstream data instead of the controller output during the last run when determining the feedback contribution(s) to the control signal. Generally speaking, such actual downstream data provides a feedback indication of the response to the control signal and thus can be measured or calculated by the downstream element (e.g., process control module) or device (e.g., actuator) receiving the control signal. Such data is provided in lieu of an implicit response to the control signal, such as the controller output from the last run. As shown in FIG. 4A and FIG. 5-FIG. 6, the continuously updated filter 116 receives the control signal 105 as an implicit indication of the downstream response. The use of such implicit data effectively assumes that the downstream element, such as the actuator, has received the communication of the control signal and is therefore responding appropriately to the control signal. Actual feedback data is also distinct from other response indications, such as measurements of the process variable being controlled.

7 depicts an exemplary controller 140 receiving actuator position data from a downstream device or element responsive to a control signal. The downstream device or element often corresponds to an actuator and provides a measurement of the actuator position. More generally, the downstream device or element may correspond to or include a PID control block, a control selector, a splitter, or any other device or element controlled by the control signal. In the exemplary case shown, the actuator position data is provided as an indication of the response to the control signal. Thus, the actuator position data is utilized by the controller 140 during continuous execution of the control routine despite the lack of measurement updates of the process variable. To this end, the continuously updated filter 116 can receive the actuator position data via a communication stack 146 that establishes an interface to the incoming feedback data. In this exemplary case, the feedback data includes two indications of the response to the control signal, the actuator position, and the process variable.

As with the previous embodiment, the continuously updated filter 116 is configured to adapt to situations involving a lack of a measurement update for the process variable. The continuously updated filter 116 similarly recalculates its output during such a lack, despite the fact that only the filter output generated after receiving a new measurement update is used by the summer 112. However, upon receiving a measurement update, the continuously updated filter 116 no longer relies on feedback of the control signal to modify its output. Rather, actual response data from the actuator is utilized as shown below.

The use of actual indications of response to control signals can help improve the accuracy of the control technique both during periods of periodic communication and during periods of non-periodic communication from the PID control element to downstream elements, such as actuators, or after communication loss. However, transmission of actual response indications typically requires additional communication between the field device and the controller when implemented in different devices. Such communication may be wireless, as described above, and therefore may be susceptible to unreliable transmission or power limitations. Other reasons may also lead to unavailability of feedback data.

As explained below, the control techniques discussed herein can also address situations where such response indications are not communicated in a periodic or timely manner. That is, application of the control techniques need not be limited to a lack of measurement updates of a process variable. Rather, the control techniques can be advantageously utilized to address situations involving a lack of other response indications, such as the position of an actuator or the output of a downstream control element. Still further, the control techniques can be utilized to address situations involving a loss, delay, or other unavailability of transmission from a controller (or control element) to a downstream element, such as a field device (e.g., an actuator) or another control element (e.g., a cascaded PID control, splitter, etc.).

Wireless or other unreliable transmission of additional data (i.e., response indications or feedback of downstream elements) to or from the controller or control element (i.e., control signals) provides additional potential for communication problems and/or challenges. As described above, feedback from downstream elements (e.g., actuators) can be involved in determining the integral contribution (or other control parameters or contributions). In this embodiment, the control routine relies on two feedback signals rather than the single process variable that is fed back in the embodiment described above. Furthermore, if the control signal has not yet reached the downstream element, the process will not benefit from the control scheme. The transmission of any one of these signals may be delayed or lost, and thus the techniques described herein address both possibilities.

A lack of response indication involving a filter or other control calculation can be addressed by maintaining an indication of the expected response (or other control signal component) until an update is received.

When the control signal does not reach the downstream element, the response indication (i.e., feedback) from the downstream element does not change. In such cases, the lack of a change in value can trigger logic in the controller (or control element) to similarly maintain an indication of the expected response (or other control signal component) until a change in value is received.

The present control techniques can also be implemented in situations where actual feedback data is not desired or available. The former case may be advantageous in situations where the simplicity of using an implicit response to a control signal is beneficial. For example, communication of actual feedback data may be problematic or impractical. The latter case may involve actuators or other devices that are not configured to provide position measurement data, as described above. Older devices may not have such capabilities.

To accommodate such devices, a switch or other device can be provided to allow the control technique to use either implicit or actual response indications. As illustrated in FIG. 8, the controller 150 is coupled to the switch 152 and thus receives both implicit and actual response indications. In this case, the controller 150 can be the same as any of the controllers described above, since the implementation of the control scheme does not depend on knowing the type of response indication. The switch 152 can be implemented in software, hardware, firmware, or any combination thereof. The control of the switch 152 can be independent of the implementation of the controller 150 and any control routines. Alternatively or additionally, the controller 150 can provide control signals to configure the switch 152. Furthermore, the switch 152 can be implemented as part of the controller itself, and in some cases can be integrated as part of the communication stack or other parts of the controller.

Practice of the control methods, systems, and techniques is not limited to any one particular wireless architecture or communication protocol. A suitable exemplary architecture and communication support scheme is described in U.S. patent application Ser. No. 11/156,215, filed Jun. 17, 2005, entitled "Wireless Architecture and Support for Wireless Communications Protocols," which is incorporated herein by reference in its entirety.
The modifications to the control routines are described in "Process Control Systems," vol. 13, No. 1, pp. 1111-1125, 2003, the disclosure of which is incorporated herein by reference in its entirety. Indeed, the modifications to the control routines are well suited to any situation in which the control routines are performed in a periodic manner but without process variable measurements for each control iteration. Other exemplary situations include those in which sampled values are provided irregularly or infrequently, for example by an analyzer or via laboratory samples.

Implementation of the present control technique is not limited to use with single-input, single-output PID control routines (including PI and PD routines), but rather can be applied to many different multiple-input and/or multiple-output control schemes, as well as cascaded control schemes. More generally, the present control technique can also be applied in the context of any closed-loop model-based control routine involving one or more process variables, one or more process inputs, or other control signals, such as model predictive control (MPC).

9 illustrates a further exemplary control system using the principles described herein, but configured with a controller 300 in the form of a speed-based controller. In the exemplary system of FIG. 9, the controller 300 operates to receive a setpoint signal, for example from one of the workstations 13 (FIGS. 1 and 3), or from any other source within or in communication with the process control system, and generate one or more control signals 305 that are provided to the process 301 from an output of the controller 300. In addition to receiving the control signals 305, the process 301 may be subject to measured or unmeasured disturbances, as shown diagrammatically in FIG. 4 by arrows 304. Depending on the type of process control application, the setpoint signal may be changed at any time during the control of the process 301, such as by a user, a tuning routine, or the like. Of course, the process control signals 305 may control actuators associated with valves or any other field devices to affect the response of the operation of the process 301. The response of the process 301 to changes in the process control signal 305 is measured or sensed by a transmitter, sensor, or other field device 306, which may correspond, for example, to any one of the transmitters 60-64 illustrated in FIG. 3. The communication link between the transmitter 306 and the controller 300 may include a wireless connection and is illustrated in FIG. 9 using a dashed line. However, the link may also be a wired communication link or other type of communication link. For purposes of discussion, the transmitter 306 is considered to be measuring the controlled process variable (i.e., the controlled variable) or a proxy variable correlated with the controlled process variable at a slow or intermittent update rate. This slow update rate may be periodic or aperiodic, and is considered to be on the same order of magnitude as the process response rate of the process dynamics associated with the controlled process variable. Thus, process variable measurements are provided once per time interval longer than the response time of the process, once per time interval similar to the process response time, or once per time interval slightly shorter than the process response time. Therefore, in some cases, this update rate can be 1/2 to 1/4 of the process response rate (the inverse of the process response time).

In a simple embodiment illustrated in FIG. 9, the controller 300 can implement a single-input, single-output closed-loop control routine, such as a PI control routine, which is one form of a PID control routine. Thus, the controller 300 includes some standard PI controller elements, including a communication stack 380, and a control signal generator, including a summing block 308, a proportional gain element 310, and a further summing block 312. The control routine 300 also includes a direct integral feedback path, including a filter 316 and an integral output switch, including a selection block 318. However, in this case, the PI controller of FIG. 9 is configured to perform position and differential control calculations for the proportional and integral components of the control signal. Thus, the controller 300 also includes a difference block 320, which is located in the proportional component calculation path, an adder 322, which is located in the integral component calculation path, and a block 324, which uses the differentially calculated control component at its input to generate the control signal 305. Generally speaking, block 324 scales or otherwise converts the change in position (velocity) control signal, as generated within controller 300, into an analog or digital signal that is sent to the control device to instruct the control device to move a certain amount in one direction or another over a certain period of time. This block 324 may send, for example, a pulse width modulated signal, a pulse signal, a digital signal indicating turn-on time, or any other signal to the control device indicating the magnitude of the change in position over time.

As illustrated in FIG. 9, an integral filter 316 is coupled to the output of the summer 322, which in turn is coupled to receive the output of the summer 312, while a block of switches 318 is coupled to the output of the filter 316 and provides the integral or reset contribution or component of the control signal generated by the controller 300 to the summing block 312.

During each iteration or operation of the controller 300, summing block 308 compares the set point signal with the most recently received process variable measurement provided within the controller 300 from the communications stack 380 to generate an error signal (e). A proportional gain element or block 310 operates on the error signal e, for example by multiplying it by a proportional gain value _Kp , to generate a proportional contribution or component of the speed control signal. A difference block 320 then determines the change in the proportional gain value since the last controller iteration by determining the difference between the current output of the gain block 310 (generated during the last or previous controller iteration) and the previous value of the gain block 310. The summing block 312 then combines the output of the change unit 320 (i.e., the proportional contribution based on speed) with the integral or reset contribution or component of the control signal generated by the integral feedback path to generate a speed control signal 326, which is provided to an output block 324.

Importantly, however, the adder 322, filter 316, and block or switch 318 in the integral feedback path of the controller 300 operate in the following manner to generate an integral or reset contribution to the control signal: Here, the adder 322 is coupled to receive the output of the adder 312 (i.e., the velocity-based control signal representing the change in position of the movable control element) during each controller iteration, and sums that value with the previous output S of the adder 322 (generated during the last iteration of the controller 300), thereby, in effect, integrating or summing the change in the output signal over a certain period of time. The new output S of the adder 322 is provided to the integral filter 316, which generates an indication of the expected process response to the control signal 305, shown as R in FIG. 9. The filter 316 provides this expected process response signal R to the switch or block 318. However, as shown in FIG. 9, the switch or block 318 samples and clamps the output of the filter 316 at the output of the switch or block 318 whenever a new process variable measurement is received, but otherwise provides a zero (0.0) value to the adder 312 as the integral control contribution during control iterations in which no new process variable measurements are available. Thus, the output of the switch 318 provided as the integral control contribution to generate a new control signal during each controller iteration is the output of the filter 316, otherwise zero (0.0), only during controller iterations in which new process variable measurements are available for use by the controller 300. Each time the new value flag generated by the communication stack 380 is set (indicating that a new process variable measurement is available to the controller 300), the adder 322 sets its output to zero and begins summing over a new period. Thus, in effect, the adder 322 sums the change in the control output signal over each controller iteration between process variable measurement changes, and resets after any new process variable measurement update is received by the controller 300 (during a controller iteration).

The expected process response to a change in the control signal, as produced by the filter 316, may be approximated using a first-order model, as described in more detail below. More generally, however, the expected process response may be generated using any suitable model of the process 301 and is not limited to a model incorporated into the feedback path of the controller 300 or to a filter or model associated with determining an integral or reset contribution to the control signal. For example, a controller utilizing a model to provide an expected process response may incorporate a derivative contribution such that the control routine 300 implements a PID control scheme. One example of incorporating an exemplary derivative contribution is described below in connection with FIG. 10.

In any event, the control technique described below allows the controller 300 to use a positive feedback path for determining the reset or integral contribution when it receives slow or non-periodic updates of the process variable, yet still allows for a robust controller response in the event of setpoint or feedforward changes occurring between receipt of new process variable measurements. Specifically, the filter 316 is configured to calculate a new indication or value of the expected process response during each or every execution of the controller 300, regardless of whether this output of the filter 316 is provided to the summing block 312 as an integral component of the control signal. As a result, the output of the filter 316 is newly regenerated during each execution cycle of the controller routine, even if only the output of the filter 316 generated immediately after (or during) the controller 300 receives a new process measurement update from the communication stack 380 is used as the integral or reset contribution of the summer 312.

Specifically, a new indication of the expected response R as produced by the filter 316 is calculated during each controller execution cycle from the current summer output S (i.e., the totalized change in the control signal output by the summer 312 since the last process variable measurement update), the indication of the expected response produced by the filter 316 during the last (i.e., immediately preceding) controller execution cycle, and the execution period of the controller. As a result, the filter 316 is described herein as being continuously updated as it is executed to generate a new estimate of the process response during each controller execution cycle. Exemplary equations that may be implemented by the continuously updated filter 316 to generate a new expected process response or filter during each controller execution cycle are set forth below.

It will be noted that the new filter output R _N is iteratively determined by adding a damping component determined by adding the difference between the summed change S _{N-1 in} the current controller output value from summer 322 and the current filter output value R _N-1 multiplied by a coefficient that depends on the reset time T _Reset and the controller execution period ΔT. By using a filter that is continuously updated in this manner, the control routine 300 is able to better determine the expected process response when calculating the integral control signal input upon receiving a new process variable measurement, thereby being more sensitive to set point changes or other feedforward disturbances that occur between the receipt of two process variable measurements. However, this calculation of the integral path prevents the control system from winding up in the presence of slowly received or intermittent process variable feedback measurements. More specifically, it will be noted that a change in the set point (without receipt of a new process measurement) immediately results in a change in the error signal at the output of summer 308, which changes the proportional contribution of the speed control signal 326 and therefore the control signal 305. As a result, summer 322 increases its output S by that amount and filter 316 then immediately begins to generate a new expected response of the process to the changed control signal, thus updating the output before the controller 300 receives a new process measurement. Then, when the controller 100 receives a new process measurement and a sample of the filter output is clamped by switch 318 to the input of summer 312 for use as the integral or reset contribution of the control signal, filter 316 is repeating, at least in part, the expected process response that reacted to or incorporated the response of the process 301 to the set point change, based on the previously transmitted control signal 305. However, to allow the error signal e generated by summer 308 to reflect changes in the process variable during the time that the process variable measurement is received at the controller 300, this integral value is only added to the control signal 326 when a new measurement is received. In controller iterations during the time that the process variable measurement is received, the integral element provided to summer 312 is set to zero. This technique prevents or helps the control system 300 from winding up. In effect, the integral element generated by filter 316 estimates the process response during the time that the subsequent process variable feedback is received (controller iteration), and if the actual process variable response was as expected, then when a new process variable measurement is received at the controller 300, the integral component sets the value generated in the proportional path to zero. If the expected response of the process differs from the actual process response during this time, the integral component changes the control signal 326 to move the actuator, thereby correcting the actuator position.

In the past, reset contribution filters used in the feedback path of non-periodically updated controllers, such as in the systems described in U.S. Patents 7,587,252 and 7,620,460, only calculated a new indication of expected response when a new process variable measurement was available. As a result, the reset contribution filter did not compensate for setpoint changes or feedforward disturbances that occurred between receipt of a process variable measurement, because the setpoint changes or feedforward disturbances were completely independent of any measurement updates. For example, if a setpoint change or feedforward disturbance occurred between two measurement updates, the expected process response of the controller could be distorted because the calculation of a new indication of expected response was based on the time since the last measurement update and the current controller output 305. As a result, the filter 316 could not begin to compensate for time changes in the process (or control signal) that resulted from a setpoint change (or other feedforward disturbance) that occurred between receipt of two process variable measurements at the controller.

However, as will be appreciated, the control routine 300 of FIG. 9 provides a predicted process response by basing the calculations made on slow or non-periodic measurements, while in addition determining a predicted response between receipt of two measurements to compensate for changes caused by set point changes (or any measured disturbances used as feedforward inputs to the controller 300). Thus, the control technique described above can accommodate set point changes, feedforward actions on measured disturbances, etc. that may affect the predicted process response, thus providing a more robust control response. Furthermore, since this control technique avoids windup in the controller, it can operate effectively when the feedback rate of the process variable measurements is equal to or even lower than the inverse of the process response time (i.e., when the time between feedback measurements being received at the controller is longer than the process response time).

As can be seen, the control technique illustrated in FIG. 9 calculates an indication of the expected response via a continuously updated filter 316 (e.g., a reset contribution filter) for each execution of the control block or routine 300. Here, the controller 300 configures the continuously updated filter 316 to calculate a new indication of the expected response for each execution of the control block. However, if the output of the filter 316 is to be used as an input to the summing block 312, the communication stack 380 and, in some embodiments, the update detection module 82 (FIG. 3) process the incoming data from the transmitter 306 to generate a new value flag for the integral output switch 318 and the adder 326 when a new process variable measurement is received. This new value flag informs the switch 318 to sample and clamp the filter output value for this controller iteration to the input of the adder 312. Alternatively, the switch 318 provides a zero (0.0) value to the adder 312 as the integral contribution value.

Whether or not a new value flag is communicated, the continuously updated filter 316 continues to calculate an indication of the expected response for each iteration of the control routine. This new indication of the expected response is conveyed to the integral output switch 318 for each execution of the control block. Depending on the presence of the new value flag, the integral output switch 318 switches between allowing the new indication of the expected response from the continuously updated filter 316 to pass to the summing block 312 or maintaining a zero value at the input to the summing block 312. More specifically, when a new value flag is communicated, the integral output switch 318 allows the immediately preceding or currently calculated indication of the expected response from the continuously updated filter 316 to pass to the summing block 312. Conversely, when the new value flag is not present, the integral output switch 318 provides a zero value to the summing block 312.

After or upon receipt of a new process variable measurement in the controller 300 and use of the output R of the continuous filter 316 in the adder 312, the time since last communication is set to zero (0) and the continuous filter output R is set to zero. Similarly, the output of the adder 322 is set to zero (0). Additionally, in these circumstances, the adder 312 may subtract the continuous filter output R from the output of the block 320 to generate a new control signal 326, depending on the manner or order in which the block 320 performs the difference calculations.

This control technique allows the continuously updated filter 316 to continue to model the expected process response regardless of whether new measurements are communicated. If the control output changes as a result of a setpoint change or a feedforward action based on a measured disturbance, the continuously updated filter 316 reflects the expected process response by calculating a new indication of the expected response at each control routine iteration, regardless of the presence of the new value flag. However, the new indication of the expected response (i.e., the reset contribution or integral component) is only incorporated into the calculation of the controller output signal when the new value flag is communicated (via the integral output switch 318), which prevents or reduces the controller winding up in response to process variable measurements received slowly at the controller 300.

Note that the simple PI controller configuration of FIG. 9 directly uses the output of the filter 316 as the reset contribution to the control signal, and in this case the reset contribution of the closed loop control routine (e.g., the continuously updated filter equation presented above) can provide an accurate representation of the process response that determines whether the process exhibits steady-state operation. However, other processes, such as processes dominated by dead time, may require incorporating additional components into the controller of FIG. 9 by including a dead time unit in the integral calculation path to model the expected process response, as illustrated in FIGS. 5 and 6. For processes that are adequately represented by a first-order model, the process time constant can generally be used to determine the reset time of the PI (or PID) controller. More specifically, if the reset time is set equal to the process time constant, the reset contribution generally overrides the proportional contribution so that the control routine 300 reflects the expected process response over time. In the embodiment illustrated in FIG. 9, the reset contribution can be achieved by a positive feedback network with a filter having a time constant the same as the process time constant. Although other models can be used, the positive feedback network, filter, or model provides a convenient mechanism for determining the expected response of a process with a known or approximated process time constant. For processes requiring PID control, the derivative contribution, also known as rate, to the PID output may also be recalculated and updated only when a new measurement is received. In such cases, the derivative calculation may use the elapsed time since the last new measurement. Although other controller components can be used to control more complex processes using non-periodic receipt of process measurements, one example of a controller that can use the filtering technique of FIG. 9 to provide robust control in response to set point changes is described below in conjunction with FIG. 10.

Specifically, FIG. 10 depicts an alternative controller (or control element) 400 that differs from the controller 300 described above in FIG. 9 in that a derivative or rate contribution component is incorporated into the controller 400. By incorporating a derivative contribution, the control routine implemented by the controller 400 includes an additional feedback mechanism, such that in some cases a proportional-integral-derivative (PID) control scheme is implemented.

The control routine or technique of Figure 10 includes a derivative contribution configured in a manner similar to that described above in connection with the systems of Figures 7 and 8 to accommodate slow, non-periodic, or otherwise unavailable updates of the process variable measurement. The derivative contribution can be reconfigured based on the time elapsed since the last measurement update. In this manner, spikes in the derivative contribution (and the resulting output signal) are avoided. More specifically, the derivative contribution in the system of Figure 10 is determined by a derivative block 432, which receives the error signal (multiplied by the proportional gain _Kp ) from the gain block 310 in parallel with elements dedicated to proportional and integral contributions, and operates to generate a derivative control component O _D , which is then provided to a change block 433. The change block 433 determines the change in the derivative control component O _D since the last controller iteration and provides this change to a summer 434 which sums or adds this change in the derivative control component to the output of the summer 312 to generate a change in the control signal. In this case, the summer 322 of the integral contribution calculation path is connected to the output of the summer 434. However, the components of FIG. 10 which are also illustrated in FIG. 9 operate in the manner described with respect to FIG. 9. Here, it will be seen that the derivative block 432 only operates to calculate a new derivative component O _D only during a controller iteration in which it receives a new value flag (indicating that a new value of the process variable measurement has been received at the controller). This operation, in effect, keeps the output of the change block 434 at zero for all controller iterations during or when no new process variable measurements are received at the controller 400.

To accommodate unreliable transmissions and, more generally, unavailability of measurement updates, the differential contribution _{O_D} is maintained at the last determined value until a measurement update is received, as indicated by a new value flag from the communication stack 380. This technique allows the control routine to continue periodic execution according to the normal or established execution rate of the control routine. Upon receiving an updated measurement, as illustrated in FIG. 10, the differential block 432 can determine the differential contribution according to the following equation:

Of course, if desired, the derivative component calculation block 432 can be connected directly to the output of the summer 308 to receive the error signal, and the derivative gain term _KD can be set to incorporate the derivative gain with the proportional gain _KP . This technique for determining the derivative contribution allows measurement updates for the process variable (i.e., the control input) to be lost or unavailable for one or more consecutive execution periods without producing output spikes, which allows for a bumpless recovery. When communication is re-established or a new process variable measurement is received at the controller, the derivative contribution equation term ( _eN - _eN-1 ) can generate the same value as it would in a standard calculation of the derivative contribution. However, for standard PID techniques, the divisor in determining the derivative contribution is the execution period. In contrast, the present control technique utilizes the elapsed time between two successfully received measurements. By using an elapsed time that is longer than the execution period, the control technique generates a smaller derivative contribution than the standard PID technique, and spiking is reduced.

To facilitate the determination of the elapsed time, the communication stack 80 can provide the new value flag described above to the derivative block 432 as shown in FIG. 10. Alternative embodiments can include or involve detection of new measurements or updates based on the value. Also, the process measurement can be used in place of the error in the calculation of the proportional or derivative components. More generally, the communication stack 380 can include or involve any software, hardware, or firmware (or any combination thereof) to implement a communication interface with the process 301, including any field devices within the process 301, process control elements external to the controller, etc. However, in the controller 400 of FIG. 10, the continuously updated filter 316 and switch module 318 operate in the same manner as described above with respect to the controller 300 of FIG. 9 to provide robust control in response to set point changes.

Practice of the control methods, systems, and techniques described herein is not limited to any one particular wireless architecture or communication protocol. A suitable exemplary architecture and communication support scheme is described in U.S. Patent Application No. 11/156,215, entitled "Wireless Architecture and Support for Process Control Systems," filed June 17, 2005, the entire disclosure of which is incorporated herein by reference. Indeed, the modifications to the control routine are well suited to any situation in which the control routine is performed in a periodic manner but without process variable measurements for each control iteration. Other exemplary situations include those in which sampled values are provided irregularly or infrequently, for example by an analyzer or via laboratory samples.

Implementation of the present control technique is not limited to use with single-input, single-output PID control routines (including PI and PD routines), but rather can be applied to many different multiple-input and/or multiple-output control schemes, as well as cascaded control schemes. More generally, the present control technique can also be applied in the context of any closed-loop model-based control routine involving one or more process variables, one or more process inputs, or other control signals, such as model predictive control (MPC).

11-14 provide graphical depictions of the simulated operation of the control routine described herein (specifically, the operation of FIG. 10) compared to a prior art controller using a standard speed-form PID control algorithm to illustrate the effectiveness of the present control routine in situations where the process response time is similar to or even shorter than the time between updates of the controlled process variable measurements. The graphs of FIGS. 11-14 illustrate simulated control examples using only primary control, although other types of control, such as override control, can be used. Generally speaking, each of the graphs of FIGS. 11A, 12A, 13A, and 14A illustrate the operation of a standard prior art speed-based PID control algorithm using both a wired feedback configuration (the operation of which is illustrated on the left side of the graph) and a wireless configuration (the operation of which is illustrated on the right side of the graph) that can utilize new process variable measurements during each controller iteration. However, Figures 11A and 13A illustrate a control situation where the feedback rate or time between process variable measurements is 8 seconds and the process response time is 8 seconds, while Figures 12A and 14A illustrate the operation of a prior art controller where the feedback rate or time between process variable measurements is 8 seconds and the process response time is 3 seconds. Similarly, Figures 11 and 12 illustrate the controller operation in response to a set point change, while Figures 13 and 14 illustrate the operation of these same controllers in response to a disturbance change. For comparison, the graphs of Figures 11B, 12B, 13B, and 14B respectively illustrate the operation of the speed-based PID algorithm described herein in the same process control situation as Figures 11A, 12A, 13A, and 14A.

Generally speaking, the following parameters were used in the simulated control operations depicted in Figures 11-14, and these tests were conducted for wired and wireless inputs, as well as for set point changes and unmeasured disturbances, as described above. The simulation of the control and process used in the tests was set up as follows:
Test for 8 second process response First order process (same gain and dynamics)
Process Gain = 1
Process time constant = 8 seconds Process dead time = 0 seconds PID tuned to first order (lambda coefficient 1.0)
Proportional gain = 1
Integral gain = 7.5 repetitions/min Test for 3 second process response First order process (same gain and dynamics)
Process Gain = 1
Process time constant = 3 seconds Process dead time = 0 seconds PID tuned to first order (lambda coefficient 1.0)
Proportional gain = 1
Integral Gain = 20 iterations/min. Module Execution Rate 0.5 sec for all tests Wireless Communication Update Rate 8 sec period for all tests Disturbance Input Affects only primary measurements Gain = 1

As illustrated in Figures 11A and 13A, the prior art speed-based PID control algorithm performs somewhat satisfactorily in response to set point changes (Figure 11A) and disturbance changes (Figure 13A) in both wired and wireless configurations where the process response time is equal to the interval between process variable measurements (both set to 8 seconds). However, as shown by the circled portions of the graphs in Figures 11A and 13A, this control technique causes significant variations in valve position in response during wireless control. As illustrated in Figures 11B and 13B, the control technique described herein performs somewhat better in these situations (set point changes in Figure 11B and disturbance changes in Figure 13B) and is very similar to the behavior of the wired configuration.

However, as illustrated in Figures 12A and 14A, the prior art speed-based PID controller does not perform very well, and in fact becomes unstable, during wireless control in response to both set point changes and disturbance changes when the process response time is 3 seconds and the process variable update rate is 8 seconds. However, as illustrated in Figures 12B and 14B, the current speed-based control routine still performs very satisfactorily in these circumstances, illustrating the effectiveness of the presently described control routine when the update interval time of the process variable measurement is greater (longer) and even significantly greater (e.g., 2-4 times) than the process response time.

The term "field device" is used broadly herein to include several devices or combinations of devices (i.e., devices that provide multiple functions, such as transmitter/actuator hybrids), as well as any other device(s) that perform a function in a control system. In any case, field devices may include, for example, input devices (e.g., devices such as sensors and instruments that provide status, measurements, or other signals indicative of process control parameters, such as temperature, pressure, flow rate, etc.), as well as control operators or actuators that perform actions in response to commands received from controllers and/or other field devices, such as valves, switches, flow control devices, etc.

It should be noted that any control routine or module described herein may have portions thereof implemented or executed in a distributed manner across multiple devices. As a result, a control routine or module may have portions implemented by different controllers, field devices (e.g., smart field devices) or other devices, or other control elements, as desired. Similarly, a control routine or module implemented within a process control system as described herein may take any form, including software, firmware, hardware, etc. Any device or element involved in providing such functionality may be referred to herein generally as a "control element," regardless of whether the software, firmware, or hardware associated therewith is located in a controller, field device, or any other device (or collection of devices) within the process control system. A control module may be any part or portion of a process control system, including, for example, a routine, block, or any element thereof stored on any computer-readable medium. Such a control module, control routine, or any portion thereof (e.g., block) may be implemented or executed by any element or device of the process control system, generally referred to herein as a control element. The control routines, which may be modules or any part of a control procedure, such as a subroutine, a portion of a subroutine (such as a sequence of code), etc., may be implemented in any desired software format, using object-oriented programming, or using ladder logic, sequential function charts, function block diagrams, or any other software programming language or design paradigm. Similarly, the control routines may be hard-coded, for example, into one or more EPROMs, EEPROMs, application specific integrated circuits (ASICs), or any other hardware or firmware elements. Still further, the control routines may be designed using any design tool, including graphical design tools, or any other type of software/hardware/firmware programming or design tool. Thus, the controller 11 may be configured to implement the control strategies or control routines in any desired manner.

Alternatively or additionally, the function blocks may be stored in and implemented by the field devices themselves or other control elements of the process control system, which may be the case if the system utilizes fieldbus devices. Although the control system descriptions are provided herein using function block control strategies, the control techniques and systems may also be implemented or designed using other conventions, such as ladder logic, sequential function charts, or any other desired programming language or paradigm.

In implementation, any of the software described herein can be stored in any computer-readable memory, such as a magnetic disk, laser disk, or other storage medium, the RAM or ROM of a computer or processor. Similarly, the software can be distributed to a user, process plant, or operator workstation using any known or desired distribution method, including, for example, a computer-readable disk or other portable computer storage mechanism, or a communication channel, such as a telephone line, the Internet, the World Wide Web, any other local area network, or a wide area network (wherein distribution is considered to be the same as or compatible with providing such software via a portable storage medium). Furthermore, the software can be provided directly without modulation or encryption, or can be modulated and/or encrypted using any suitable modulation carrier and/or encryption technique before being transmitted over a communication channel.

Thus, the present invention has been described with reference to specific embodiments that are intended to be illustrative only and not limiting, but it will be apparent to one skilled in the art that modifications, additions, or deletions can be made to the control techniques without departing from the spirit and scope of the invention.

Claims (18)

a process controller generating a control signal for controlling a process variable during each of a number of controller iterations of the process controller;
The process controller includes:
a communications unit for receiving new values of the process variables at a rate that is less than a rate of the multiple controller iterations of the process controller;
a proportional control element that generates a proportional control signal value during each of the iterations of the process controller,
a proportional control element including: a first summer to determine a difference between a set point value for the process variable and a received value of the process variable; and a proportional gain unit coupled to the first summer;
an integral control element that generates an integral control signal value during each of the iterations of the process controller;
a continuously updated iterative filter that determines the integral control contribution during each iteration of the process controller based on a previous value of the integral control contribution generated during a previous iteration of the process controller and based on a total control signal value for all previous iterations since the last iteration in which a new process variable measurement was received ;
a switch coupled to the continuously updated repetitive filter that receives the integral control contribution, the switch being operative to (1) provide the integral control contribution generated by the continuously updated repetitive filter as the integral control signal value only during process controller iterations associated with receipt of a new measurement of the process variable at the communication unit, and (2) provide a previous value of the integral control contribution or a zero value as the integral control signal value during process controller iterations not associated with receipt of a new value of the process variable received at the communication unit ; and
and an integral control element including a second summer that sums the proportional control signal value and the integral control signal value during each process controller iteration to generate the control signal;
the integral control element further includes a third adder coupled to the continuously updated repetitive filter, the third adder summing the control signals for previous controller iterations to generate the total control signal value , the third adder providing the total control signal value as an input to the continuously updated repetitive filter;
The process controller, wherein the third adder resets the total control signal value upon receiving a new value of the process variable at the communication unit. 2. The process controller of claim 1, wherein the third adder resets the total control signal value to zero upon receiving a new value of the process variable at the communication unit. The process controller of claim 1 or 2, wherein the proportional control element includes a difference unit coupled to the proportional gain unit. 4. The process controller of claim 3, wherein the difference unit is coupled between the proportional gain unit and the second summer to determine a difference value between an output of the proportional gain unit from a previous controller iteration and an output of the proportional gain unit at the current controller iteration, and provides the difference value to the second summer as the proportional control signal value. The process controller of any one of claims 1 to 4, further comprising a control signal conversion unit coupled to the second adder that converts the control signal into an output control signal that is transmitted to control a device in a process. 6. The process controller of claim 1, further comprising a derivative control element for determining a derivative control signal value, the derivative control element comprising a derivative gain unit coupled to the first summer and a second difference unit coupled to the derivative gain unit for determining a further difference value between an output of the derivative gain unit from a previous controller iteration and an output of the derivative gain unit at the current controller iteration and providing the further difference value to the second summer as the derivative control signal value, the second summer summing the derivative control signal value with the integral control signal value and the proportional control signal value to generate the control signal. The process controller of any one of claims 1 to 6, wherein the previous value is an integral control signal value output by the switch in a previous controller iteration. 1. A method for generating process control signals during each of multiple iterations of a control routine, comprising:
receiving, via a computing device, new values for a controlled process variable at a rate that is less than a rate of the multiple controller iterations of the control routine;
using a computer processing device to generate a difference-based integral feedback contribution for use in generating a control signal during each of the multiple iterations of the control routine, the difference-based integral feedback contribution including summing all previous iterations of the control routine since a last iteration of the control routine that received a new process variable value to generate a total control signal value , providing the total control signal value to a continuously updated iteration filter, and using the continuously updated iteration filter during each of the multiple iterations to determine a current integral feedback contribution value for a current iteration of the control routine from the integral feedback contribution values of prior iterations of the control routine and the total control signal value;
(1) using the current integral feedback contribution value determined by a switch coupled to the continuously updated iterative filter to generate the control signal for the current iteration of the control routine only during each iteration that receives a new process variable value, and (2) not using the current integral feedback contribution value determined by the switch or a zero value to generate the control signal during an iteration that does not receive a new process variable value from a process;
using the control signal to control the process variable;
said summing includes generating said total control signal value using an adder that sums said control signals for previous controller iterations, said adder providing said total control signal value as an input to said continuously updated iterative filter;
resetting the sum control signal value when the adder receives a new value of the process variable;
A method for generating a process control signal. The method of generating a process control signal of claim 8, further comprising: using a computer processing device to generate a proportional contribution based on the difference during each of the iterations of the control routine; and using the proportional contribution during each of the iterations of the control routine to generate the control signal. The method for generating a process control signal according to claim 8 or claim 9, wherein the new process variable value is a measurement of a process parameter that is influenced by the control signal. The method of generating a process control signal as described in any one of claims 8 to 10, wherein determining the integral feedback contribution comprises generating the current integral feedback contribution value based on a difference between the total control signal value for the current iteration of the control routine and the integral feedback contribution value of the previous iteration of the control routine multiplied by a coefficient that depends on a reset time and a controller execution period. The method of generating a process control signal according to any one of claims 8 to 11, wherein not using the current integral feedback contribution value to generate the control signal during an iteration in which no new process variable values are received from the process comprises using a fixed value as the current integral feedback contribution to generate the control signal during an iteration in which no new process variable values are received from the process. The method for generating a process control signal according to claim 12, wherein the fixed value is zero. a process controller generating a control signal for controlling a process variable during each of a number of controller iterations of the process controller;
The process controller includes:
a communications unit for receiving new values of the process variables at a rate that is less than a rate of the multiple controller iterations of the process controller;
an integral control element that generates an integral control signal value during each of the iterations of the process controller;
a continuously updated iterative filter that determines the integral control contribution during each iteration of the process controller based on a previous value of the integral control contribution generated during a previous iteration of the process controller and based on a total control signal value for all previous iterations since the last iteration in which a new process variable measurement was received ;
a switch coupled to the continuously updated repetitive filter that receives the integral control contribution, the switch (1) operative to provide the integral control contribution generated by the continuously updated repetitive filter as the integral control signal value only during process controller iterations associated with receipt of a new measurement of the process variable at the communication unit, and (2) providing a previous value or a zero value of the integral control contribution as the integral control signal value during process controller iterations not associated with receipt of a new value of the process variable at the communication unit; and
an integral control element including a control signal generator coupled to receive the integral control signal value to generate a control signal for use in controlling a process during each of the controller iterations;
the integral control element further includes an adder coupled to the continuously updated repetitive filter, the adder summing the control signals for previous controller iterations to generate a total control signal value, the adder providing the total control signal value as an input to the continuously updated repetitive filter;
The process controller, wherein the adder resets the total control signal value upon receiving a new value of the process variable at the communication unit. a proportional control element that generates a proportional control signal value during each of the iterations of the process controller,
a proportional control element including: a first summer to determine a difference between a setpoint value for the process variable and a received value of the process variable; and a proportional gain unit coupled to the first summer;
15. The process controller of claim 14, wherein the control signal generator includes a second summer that sums the proportional control signal value and the integral control signal value during each process controller iteration to generate the control signal. The process controller of claim 15, wherein the proportional control element includes a difference unit coupled to the proportional gain unit. 17. The process controller of claim 16, wherein the difference unit is coupled between the proportional gain unit and the second summer to determine a difference value between an output of the proportional gain unit from a previous controller iteration and an output of the proportional gain unit at the current controller iteration, and provides the difference value to the second summer as the proportional control signal value. 18. The process controller of claim 17, further comprising a derivative control element for determining a derivative control signal value, the derivative control element comprising a derivative gain unit coupled to the first summer and a second difference unit coupled to the derivative gain unit for determining a further difference value between an output of the derivative gain unit from a previous controller iteration and an output of the derivative gain unit at a current controller iteration and providing the further difference value to the second summer as the derivative control signal value, the second summer summing the derivative control signal value with the integral control signal value and the proportional control signal value to generate the control signal.