JPS5936803A - Method for correcting feedforward model - Google Patents

Abstract

PURPOSE:To improve the quality of process control by correcting a factor or a bias intermittently so that an operation signal and an output signal from a static characteristic compensating feedforward model coincide with each other at the time of stable state reducing external disturbance. CONSTITUTION:A static characteristic compensating signal Dn being an output from the feedforward model 38, is applied to a differential part 51 to find the difference and turns the signal to a speed type. The signal Dn is successively inputted to an adder 33 through a contact 62b, so that the signal Dn is added with an output from a control part 32 and returned to a position type by a conversion part 34 to form a signal A. On the other hand, the signal Dn is added with the signal A by an adder 35 through an imperfect differentiation part 42 and a contact 62b'. The signal A is applied to a subtractor 43 to find the difference between the signal A and the signal Dn, the differential signal is integrated by an integration part 44 through a contact 62a and the integrated signal is applied to a multiplying part 40. The factor of the multiplying part 40 is adjusted so that the integrated value becomes zero to correct the feedforward model 38 so that the signal A and the static characteristic compensating part Dn after correction coincide with each other.

Description

[Detailed Description of the Invention] [Technical Field of the Invention] The present invention relates to a method for automatically correcting a feedforward model in response to changes in disturbance characteristics in a process control device that combines feedback control and feedforward control. It is.

[Technical background of the invention]

Feedback control plays an important role in process control, but this control method focuses only on results, and performs corrective control when the results do not match the target. Therefore, there is no problem in the case of slow fluctuations, but the results deviate from the target value even in the case of sudden disturbances. A major drawback is that the response is delayed. Therefore, we aim to improve controllability by applying feedforward control in combination with feedback control, which measures the disturbance and proactively compensates for the effect of the disturbance before it appears on the control amount. .

FIG. 1 is a block diagram showing a basic system that combines feedback control and feedforward control. At the same figure pressure, the comparator 1 compares the target value and the control amount, and the PIDID adjustment button 2 determines the deviation, and the trowel performs the adjustment calculation.
It leads to the adding section 3. on the other hand.

The disturbance affects the control amount X through a disturbance transmission path 7 having a transfer function GD. Therefore, in order to cancel and compensate for this influence, the disturbance compensation signal is sent to the adder 3 via the feedforward model 8 having the transfer function GF.
This is added to the output signal of the adjustment section 2, and this is added to the controlled object 7 as an operation signal to control it, and the transfer function GP of this controlled object is determined.
The control amount X is the sum of the output of the block 4 having the value and the influence of the disturbance (this addition is expressed as being performed by the adding section 5).

Therefore, if the output signal of controller 2 is Y.

X=4Y+1)XUF)XGP+DXGD=YXGP+
DX (GD 1 GFXGP) ...... 11) (
From equation 1), no matter how the disturbance changes, the control amount XK
For the first time that has no influence, G l) - F G ), X GP = 0 must hold. Accordingly, the transfer function GF of the feed forward model 8 needs to be determined as follows. GDI G
, can generally be approximated by a combination of dead time and first-order delay.

Here, K, , K,) is the gain number, Tp * T
l) is Tokisada Akira, I, P, LD is Muda F1 Kawabo. Therefore, the history of feedforward model 8: +4 cabinet number G, becomes... (3). Also, if the delay times of G and GD are equal to y, then the equation becomes as follows, which is a simple formula. For practical purposes, this simple formula is sufficient. Conventionally, the coefficient KD of the transfer function GD with respect to the disturbance of the controlled object is constant, and the coefficient KF=KD/KP of the feedforward model is also constant. However, in the actual σ) process, the disturbance coefficient KD of C does not have a fixed value, but includes indirect disturbance elements, changes in characteristics over time, changes in internal and external physical quantities, changes in chemical components, changes in ambient temperature, and changes in the measured temperature. It changes irregularly and significantly due to the influence of external disturbance elements. CO) f, -me There is a problem in that the effect of feedforward control is lost, or on the contrary, it causes harm.

Especially recently, the raw materials and fuel used in Zorocess are increasing (1
), product diversification, load changes due to changes in operating rates due to changes in the economic environment, multi-purpose use, etc., there is a strong need to increase the flexibility of processes and, therefore, the flexibility of control devices. It has become.

The above points will be explained in more detail below, taking heat exchanger mouth temperature control as an example. In FIG. 2, a raw material 11 is led through a pipe 12 to a heat exchanger 14, where it is heated with steam and taken out. The outlet temperature T of the heat exchanger 14. is detected by the temperature detector 15 and entered into the temperature controller 19, and the outlet temperature T is determined. Adjust and calculate so that it becomes a predetermined value ◎ On the other hand, the raw material mud amount Fi is detected by the raw material flow rate detector 13,
This is input into the feedforward model 21, and its output is added to the output of the controller 19 on the adder side, and the increase result is given to the steam flow rate controller 4 as a set value. In the steam kick regulator n, the steam flow control signal detected by the steam flow rate detector 17 is used as a feedback signal to perform adjustment calculations so as to be equal to the upper Re setting value, and the output No. 48 is used to control the control valve 18. Control the opening degree. In this way, heat exchanger 1
4 outlet temperature T.

control to keep it constant.

Transfer function G of feedforward model 22 in the above example
Consider F. First, the heat balance Q in the steady state of the process is determined.

■ Q"FsXHs=HXFIXCHX(TS'ri)"
(5) Here, F8 is the steam weight flow rate, H8 is the steam latent heat, Fi is the raw material car l, Ci is the specific heat of the raw material, Ts is the heat exchanger outlet temperature setting temperature, and Ti is the heat exchanger inlet temperature of the raw material , η is the heat exchanger efficiency.

The steam flow rate F8, which is the manipulated variable, is calculated from equation (5).

becomes. From equation (6), the static characteristic compensation component GFS of the transfer function of the feedforward model 22 is as follows, and the transfer function GF of the feedforward model 4 to which the dynamic characteristic compensation component is added is as follows. Here, TD is the time constant from the raw material flow rate detector 13 to the outlet temperature detector 15, and TP is the steam flow rate setting (adjustment
This is the time constant at the trench where this effect reaches the heat exchanger outlet temperature.

In the above example, only the change in the raw material flow rate Fi is assumed to be a disturbance, and the coefficient ζ is constant θ, but in reality, the following factors, that is, (1 ) Original [Change in temperature (11) Change in heat exchange efficiency (ill) Change in steam retention, W heat exchange (1) Change in ambient temperature/ratio M Due to changes in specific heat of JiC material, KF is irregular As a result, the effect of feed-to-word control is not fully demonstrated, and
Controllability when changing the raw material flow rate deteriorates, and product quality is disturbed.

As described above, conventional control methods have the problem that feedforward control is not effective and may have the opposite effect in some cases, and this problem is becoming more severe as the demand for process flexibility becomes stronger. It had become something big.

[Purpose of the invention]

An object of the present invention is to provide a method for automatically correcting a feedforward model that can constantly maintain feedforward compensation in an optimal state and improve the quality of process control.

[Summary of the invention]

The present invention can be used to adjust the operating end in a stable state with little disturbance, or when the operating signal used as a setting signal for the secondary adjustment loop and the static characteristic compensation calculated by the feedforward model match each other. The coefficient straddle bias of the feedforward model was weighted down as shown in the figure.

[Embodiments of the invention]

A Kazuo embodiment of the present invention will be described below with reference to FIG. In this embodiment, the #4 node output signal of the foot knock control system is of the speed type.

The process variable Pv obtained by measuring the control system
lead to. The adjustment section 32 has P (proportional), ■ (integral), D (
(differential) operation, or a combination of adjustment operations. The output signal 32a of the adjustment section 32 is applied to the controlled object 36 as an operation signal M via an addition section 33, a velocity type-position shadow signal conversion section 34, and an addition section 35. In this way, a feedback control system is configured that controls by feeding back the control lx.

On the other hand, for the feed 7 source control, the disturbance signal is input to the calculation section 39, where calculations are performed to obtain the static characteristic compensation. The output of the calculation unit 4o is sent to the multiplication unit 40, where it is multiplied by a coefficient to produce the static characteristic compensation signal Dn.
becomes. In this way, the static characteristic compensation feedforward model 38 is formed by the calculation section 39 and the multiplication section 40.

That is, when formula (4) is modified, the first term on the right side is the static characteristic compensation component, which is sought in the #characteristic compensation feedforward model 38. On the other hand, the second term on the right-hand side above is dynamic/i
This is the static characteristic compensation component multiplied by the static characteristic compensation component. Therefore, the dynamic characteristic compensation component is obtained by inputting the output of the static characteristic compensation feedforward model to an incomplete differentiator 42 (to be described later) having a transfer function expressed by the equation (uo).

The static characteristic compensation signal Dn is a 1-position type-velocity type vibration converter.
・The difference unit 51 forming one component calculates the difference between the static characteristic compensation D11 at each sampling time and the static characteristic compensation Dn-1 at the previous sampling time, that is, calculates the static characteristic as shown in C. The compensation amount is converted into velocity form. Thereafter, the signal is applied to the adder 33 via the contact 62b, where it is added to the adjustment output signal 32a, and then returned to the position form by the converter 34, resulting in signal A.

Silence! The samurai compensation portion Dn is also the incomplete differential portion 42. Contact 6
2b', and is added to the signal A in the adder 35.

On the other hand, the subtractor 43 calculates the difference between the signal A from the adder 33 and the static characteristic compensation Dn, and this difference is calculated at the contact 62a.
While the contact point 62a is closed, the above difference is integrated by the integrating section 44, and the integral value 44a is added to the multiplying section 40. Then, the coefficient of the multiplier 40 is adjusted so that the above integral value becomes zero. In this way 1. Signal A and static characteristic compensation after correction. The feedforward model 38 is modified so that the values match. , the determination unit 54 closes the contact 62a to perform correction when the control system is in a stable state, and also closes the contact 62a and causes the contact 62b, 6 to perform correction.
2b' is to be opened. Whether or not the tn system is in a stable state is determined as follows.

That is, first, the output signal of the subtraction section 43 is sent to the signal level detector 5.
5, and when this signal falls outside the predetermined range, the output of the level detector 55 is turned on and the signal is input to the timer 56. When a predetermined set time has elapsed from this time, the timer 56
The output signal of is turned on. The output of the timer group is the logical product section 6
1 is input.

On the other hand, the static characteristic compensation change output from the conversion section 51 is input to the signal level detection section 57, and when it is within a predetermined range, output No. 11 of the detection section 57 is turned on. In addition, the output signal of the incomplete differentiator 42 is sent to the signal level detector 5.
8, and when the output signal of the incomplete portion 42 is within a predetermined range, the detector output signal No. 1 is turned on. Further, the deviation signal outputted from the comparator 31 is guided to a four-signal bell detection group 59, and when the deviation is within a predetermined range, the output signal of the detection section 59 is turned on. Detection units 57, 5
Output No. 1 of 8 and 59 is led to the logical product mark. Therefore, the output signal of the AND section ω is turned on and applied to the AND section 61 when the change in the disturbance is within a predetermined range, that is, when the change in the disturbance is small.

Theory II The output of the JJt section 61 is used as a set signal for the memory 62. Therefore, there is a continuous deviation of more than a predetermined value between the static time 1 main compensation part of the disturbance compensation and the operation signal (
The memory 62 is set by the output of the AND section 61 when the output of the timer 56) and the change in the disturbance is small (the output of the logic 7@ section). When set, the output of the memory 62 is turned on, its contact 62a becomes conductive, and its contacts 62b and 62b' become non-conductive. As a result,
The output (deviation) of the reduction n section 43 is supplied to the integration section 44.
The rent will be divided. The output of the integrating section 44 is led to the riding section 40,
Modify the coefficients of the feedforward model for disturbance compensationLE-
1- Used for on the other hand.

During this time, is the addition from the conversion unit 51 and the incomplete differentiation unit 42? @S
:33.

The Ape signal is supplied to the non-conducting contacts 52b and 62.
b', and the feedforward of the disturbance is prohibited.

In this way, the feedforward model Q coefficient is automatically corrected, and when the deviation is within a predetermined value, the output signal of the signal level detector turns off, and at that moment, the timer 56
The output signal of the AND section 61 also turns off. On the other hand, the output of the signal inverter 63 is turned on,
When the set time of the timer 64 has elapsed, the output of the Kuimaro 4 is turned on and the memory 62 is reset. As a result, the contact 62a becomes non-conductive, and correction of the coefficients of the feedforward model (correction of the coefficients of the multiplier 40) is stopped.On the other hand.

The contacts 62b and 62b' become conductive, and feedforward control is restarted.

As described above, the feedforward model is automatically corrected.

In the above embodiment, the adjustment output signal 3 is a combination of velocity type, but in the case of position type, the velocity type-position shadow signal conversion part A and the difference part 51 in FIG. 33 may be used as is as the housing signal A, and the static characteristic compensation valve Dn may be directly applied to the detailing section 33 via the contact 62b.

In addition, in the example of the superior, the control system is assumed to be in a stable control state with few disturbances when the control ++11 bias Men, the change in the proportional compensation valve, and the phase compensation are each smaller than a predetermined value, and the feedforward model is then corrected. However, as shown in FIG. 4, the output signal of the converter 14 may be incompletely differentiated by an incomplete differentiator 35, and when this is smaller than a predetermined value, the control may be considered to be in a stable state and the correction may be made. .

In each of the embodiments described above, each component of the control system may be of an analog type or a digital type, and may be formed by one or more individual parts, and may be programmed and formed by one computer. Good too.

Further, in each of the above embodiments, not only static characteristic compensation but also dynamic characteristic compensation is performed by feedforward control, but the present invention can also be applied to a case where only static characteristic compensation is performed. In this case, the incomplete differential part 4 in FIG.
2. The adder 35 and the like may be removed.

Furthermore, instead of correcting the coefficients of the feedforward model, the noise coefficient may be corrected. Since the coefficients or biases of the feedforward model are intermittently corrected so that the output signal of the static characteristic compensation feedforward model and the output signal of the static characteristic compensation feedforward model match each other, it is possible to maintain the feedforward compensation in an optimal state. The quality of process control can be improved.

Furthermore, since disturbance elements that vary slowly over time are automatically corrected even if they are not detected, there is an advantage that fewer disturbance elements need to be detected, and the system can be constructed easily and at low cost.

[Brief explanation of drawings]

FIG. 4 is a block diagram showing a control device according to another embodiment of the present invention. 7) It is a block diagram showing part of the i+1 JII11 device used in the implementation. 12...Adjustment unit, 14...Velocity shadow signal-position signal conversion unit, 16...Controlled object, 19...Calculation unit, 2
0... Multiplication section. 21...Position shadow signal-velocity type signal converter, 22...
Incomplete differential part, 23... Nose reduction part, 24... Integral part.

Claims (1)

[Claims] (1) A process control device in which a feedforward control mechanism for proactively suppressing the influence of disturbance is added to a feedback control system, which detects an element that becomes a disturbance and suppresses the disturbance. The static characteristic compensation signal calculated by the model or the signal obtained by converting the C error is added to the adjustment output signal of the foot control control system, and the addition result or The Iro number converted from this, or a signal determined based on these, is used as an operating signal to adjust the operating end, or as a setting signal for the secondary adjustment loop, and when in a stable state with little disturbance, the operating signal and The above output from the feedforward model? 7? 5. A method for automatically correcting a feedforward model, characterized in that coefficients or nominals of the feedforward model are intermittently corrected so that characteristic compensation components and characteristic compensation components coincide with each other. 2. In the method according to claim 1. When the adjustment output signal of the feedback control system is of the position type, the output signal of the static characteristic compensation feedforward model is directly added to the adjustment output signal of the feedback control system, and the result is used as the operation signal. When the output signal is a velocity type signal, the output No. 16 of the static characteristic Ura feedforward model is once converted into a speed type signal, and then added to the adjustment output signal of the feedback control system, and the addition result is converted into a position type signal No. 4rR. A method characterized in that a signal converted into is used as an operation signal. 3. In the method according to claim 1. The output of the static characteristic compensation feedforward model is incompletely differentiated, and the incompletely differentiated value is converted into the sum of the static characteristic compensation component and the adjustment output signal of the foot control system, or a signal obtained by converting this. A method characterized by adding the results and using the addition result as the operation signal.