JPH07181292A - Plant operation support system - Google Patents

Abstract

PURPOSE:To provide a plant operation support system for indicating information concerning the degree of normality and the like of monitor signals referred for judging by the system. CONSTITUTION:These are provided an input processor 3 for outputting monitor signals, a characteristic memory part 4 having stored a component model quantitatively simulating the static characteristic of the constituting component of the plant, a state estimation part 5 to estimate the process state of the plant form the monitor signals of the input processor 3 by using the component model is the characteristics memory part 4, a monitor signal diagnosing part 6 to compare the process state estimated in the state estimation part 5 and the monitored signals and judge the degree of normality of the monitor signals with a fuzzy inference, and a state indicator 7 to indicate the monitor signals and the judgement results of the monitor signal diagnosing part 6.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a plant operation support device for assisting the operation of a plant by automatically diagnosing a plant state, and in particular, it can automatically diagnose the soundness of various measuring instruments including sensors, and The present invention relates to a plant operation support device in which a system exhibiting chaotic behavior is also a diagnosis target.

[0002]

2. Description of the Related Art Generally, in a large-scale plant such as a nuclear power plant, when some kind of failure occurs, the abnormality does not spread over a large range, or the soundness of the equipment and facilities of the plant is maintained. As described above, a plant interlock operated by logic such as a relay provided in the plant is provided. When an abnormal event occurs in such a plant, the operator must promptly judge the state of the plant and take appropriate measures. When the scale of the abnormal event is large or the propagation of the transient phenomenon is fast, the burden on the operator becomes large.

On the other hand, the outputs of various instruments of the plant are monitored, and when an abnormal event occurs, the abnormal event is analyzed and the plant state is presented to an operator so that the plant state can be easily grasped. The proposed plant operation support device was proposed.

As such a plant operation support device, an equipment model that constitutes the plant is prepared, and the abnormality is detected early by comparing the process state of the plant estimated by the model with the actual observation signal. There are also those that try to detect anomalies early by monitoring the feature quantities in the frequency domain such as the power spectrum of noise and coherence contained in the observed signal of the plant, and their changes.

[0005]

However, in the conventional plant operation support device for early detection of abnormalities by handling the quantitative model such as the above equipment model, how much the operator can trust the determination result of the device is determined. There was no way to judge. That is, in the conventional device, there is room for ambiguity in the threshold value for determining whether or not the quantitatively calculated process state is normal, and when there is noise in the observed signal, the observed signal itself There is also ambiguity, and the operator could not know the ambiguity of the diagnostic result including such ambiguity. In addition, there is a limit to accurately estimating the state with a quantitative model during a transition, and it is difficult to perform an accurate diagnosis during a transition.

Further, the conventional plant operation support device for observing the characteristic amount of the noise contained in the observed signal in the frequency domain and making the diagnosis is constituted by a linear superposition of sinusoidal vibrations with many noises. Since it is based on the assumption, it was not possible to accurately grasp the characteristics of irregular signals due to the nonlinearity inherent in physical phenomena. With such a device, accurate diagnosis cannot be expected when the system to be monitored exhibits chaotic behavior or chaotic behavior.

Therefore, an object of the present invention is to solve the above-mentioned problems of the conventional plant operation support device, present information to the operator about the normality of the observed signal, etc., which is the basis for the device to make a decision, and thus the plant state. It is to provide a plant operation support device that contributes to an accurate grasp of

Another object of the present invention is to provide a plant operation support apparatus which also diagnoses a system exhibiting chaotic behavior.

[0009]

A first plant operation support apparatus according to the present invention inputs an output of a plant instrument,
An input processing unit that outputs an observation signal, a characteristic storage unit that stores a device model that quantitatively simulates the static characteristics of plant component equipment, and an observation signal of the input processing unit using the device model of the characteristic storage unit The state estimation unit that estimates the process state of the plant, the process state estimated by the state estimation unit and the observation signal are compared, and the observation signal diagnosis unit that determines the normality of the observation signal by fuzzy inference, and the observation signal It is characterized in that it is provided with a status display section for displaying the determination result of the observation signal diagnostic section.

A second plant operation support apparatus according to the present invention is the above-mentioned first plant operation support apparatus, in which an observation signal determined to be normal by an observation signal diagnosis unit is input to the equipment model, and the equipment model is subjected to fuzzy inference. A device characteristic diagnosis unit that determines the degree of change in the characteristic of the device described in, and the state display unit displays the observation signal, the determination result of the observation signal diagnosis unit, and the determination result of the device characteristic diagnosis unit. It is characterized by being configured in.

The third plant operation support apparatus according to the present invention is the second plant operation support apparatus according to the second plant operation support apparatus, wherein the observation signal diagnosis unit determines that the observation signal is abnormal in the input / output relationship of the predetermined constituent devices, and When the device characteristics diagnosis unit determines that the device characteristics have not changed,
A normal signal identifying unit that identifies by fuzzy reasoning the normality of the observed signal that has been determined to be abnormal by the device model of the constituent device is provided, and the status display unit is configured to display the normal value of the observed signal. It is characterized by that.

A fourth plant operation support apparatus according to the present invention is characterized in that, in the first plant operation support apparatus, the characteristic storage unit has a characteristic of chaos including a Lyapunov exponent of an observation signal showing a chaotic behavior during normal operation. The plant state changes from the plant state during normal operation by comparing the characteristic amount of chaos stored in the characteristic storage unit with the characteristic amount of chaos of the observed signal. The state display unit is configured to display the determination result of the precise diagnosis unit.

According to a fifth plant operation support apparatus of the present invention, in the above first plant operation support apparatus, the characteristic storage unit stores a qualitative model representing a dynamic characteristic of an observed signal. It is equipped with a transient diagnosis unit that compares the behavior of the observed signal estimated by the qualitative model with the behavior of the actual observed signal to determine whether the observed signal is normal. It is characterized in that the determination result is displayed.

A sixth plant operation support apparatus according to the present invention is an input processing unit for inputting an output of an instrument of a plant and outputting an observation signal, an equipment model of static characteristics of the observation signal and a qualitative model of dynamic characteristics. A characteristic storage unit that stores the, a plant state determination unit that determines whether the plant is in a steady state or a transient state based on the observation signal of the input processing unit, and a plant state determination unit that determines whether the plant is in a steady state. An observation signal that estimates the process state of the plant from the observation signal using a model and the observation signal that compares the process state estimated by the state estimation unit with the observation signal to determine the normality of the observation signal by fuzzy inference Behavior of the observed signal and actual observed signal estimated by the qualitative model of the characteristic storage unit when the diagnostic unit and the plant state determination unit determine the transient state And a state diagnosis section for displaying the observation signal, the observation signal diagnosis section, and the judgment result of the transition time diagnosis section. It is characterized by that.

In a seventh plant operation support apparatus according to the present invention, the characteristic storage unit has an observation signal including a Lyapunov exponent indicating a chaotic behavior in each operation mode including a starting operation state, an output increasing operation state and a rated operation state. It stores the characteristic amount of chaos of, and compares the characteristic amount of chaos corresponding to the operation mode with the characteristic amount of chaos of the observed signal based on the determination result of the plant state determination unit. , Is provided with a precision diagnosis unit for determining a change from a standard plant state in the corresponding operation mode, and the state display unit is configured to display a determination result of the precision diagnosis unit. .

[0016]

According to the first plant operation support apparatus of the present invention, the state estimation unit inputs the observation signal from the input processing unit and the device model of the characteristic storage unit, and outputs from the input to the predetermined device. And the observed signal diagnostic section determines the normality of the actual observed value based on the calculated output of the device. From the normality of the observed value, the operator can grasp the reliability of the observed value, that is, the soundness of the instrument as a numerical value, and judge the essence of the abnormal event.

In the second plant operation support apparatus according to the present invention, the soundness of the input / output characteristic model of a predetermined device is calculated and displayed by the device characteristic diagnosis unit in addition to the normality of the observed signal. When the characteristics of the equipment are changing with the passage of time, the operator can grasp the change, and thereby more accurately judge the plant state.

A third plant operation support apparatus according to the present invention is provided with a normal signal identifying section in addition to the configuration of the second plant operation support apparatus, and the apparatus characteristic diagnosis section judges that the characteristic model of the apparatus is correct. Then, the normal signal identifying unit calculates the degree of normality of the observed value using this characteristic model. This normal observation value allows the operator to easily understand the original plant state.

In a fourth plant operation support apparatus according to the present invention, in addition to the configuration of the first plant operation support apparatus described above, a characteristic storage section stores a characteristic amount of chaos of an observation signal, and a precise diagnosis section is provided. Since this is compared and monitored with the characteristic amount of chaos of the observed signal, a chaotic system can be a diagnostic target.

In a fifth plant operation support apparatus according to the present invention, in addition to the configuration of the first plant operation support apparatus, a characteristic storage unit stores a qualitative model representing a dynamic characteristic of an observed signal, Since the diagnosis unit makes a judgment by comparing the observed signal at the time of transient and the above qualitative model, the plant event at the time of transient can be the target of diagnosis.

The sixth plant operation support apparatus according to the present invention has a state determination section for determining whether the plant is in a steady state or a transient state based on the observation signal of the input processing section, and has a state estimation section and an observation signal diagnosis section. Since the observed signal of the steady state plant state can be evaluated by and, and the transient state plant state can be evaluated by the transient diagnosis unit, the observed signal can be automatically evaluated for any plant state. .

In the seventh plant operation support apparatus according to the present invention, in addition to the configuration of the sixth plant operation support apparatus, the characteristic storage unit stores the characteristic amount of chaos of the observation signal, and the precise diagnosis unit Compares the stored chaos characteristic quantity with the chaos characteristic quantity of the observed signal, so that the observed signal is evaluated for any plant state, and
It is possible to provide a plant operation support device capable of diagnosing a chaotic system.

[0023]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows the configuration of the first embodiment of the present invention. In FIG. 1, a plant operation support device 1 is an input processing unit 3 that inputs an output of an instrument installed in a plant 2 and outputs an observation signal, and a device that quantitatively simulates static characteristics of a component device of the plant 2. Characteristic storage unit 4 storing the model
And the observation signal from the input processing unit 3 and the equipment model from the characteristic storage unit 4, and the state estimation unit 5 that estimates the process state from the observation signal using the equipment model, and the output of the state estimation unit 5 and the actual observation. An observation signal diagnostic unit 6 that compares the signal and determines the degree of normality of the observation signal by fuzzy inference;
It is composed of an observation signal and a status display unit 7 that displays the determination result of the observation signal diagnosis unit 6.

As described above, the observation signal diagnosis unit 6 compares the process state estimated by the state estimation unit 5 with the observation signal and determines by fuzzy inference whether or not the observation signal is normal. Make a health diagnosis. This diagnosis is based on the premise that the sensor is installed so that the process state S estimated from the observed signal is mixed with redundancy. In other words, if there is no failure in all the sensors, the process state estimated from each observed signal should converge to the true state.If one sensor fails, the estimated process state has an error. It occurs, and by comparing it with the process state estimated from the observed signals of other sensors, it is possible to identify which sensor has failed.

The processing in the observation signal diagnostic section 6 will be mainly described with reference to FIGS. 2 and 3.

FIG. 2 shows the flow of processing in this embodiment. The first step 101 is processing in the input processing unit 3, which periodically takes in observed signals from the plant 2.

The next step 102 is processing in the state estimating unit 5, which uses the equipment model stored in the characteristic storage unit 4 to estimate the process state of the plant from the observed signal. For example, assume that the following nine device model formulas are prepared.

T = F11 (P, F), P = F12 (T,
F), F = F13 (P, T) P = F21 (N, F), N = F22 (P, F), F = F23
(P, N) F = F31 (T, N), T = F32 (F, N), N = F33
(T, F) where T, P, F, N are process states, F11, F1
It is assumed that 2, F13, F21, F22, F23, F31, F32, and F33 are functions that output a process state from an input signal.

F1i, F2i and F3i (i = 1, 2, 3) are independent of each other, and Fi1, Fi2 and Fi3 (i = 1, 1)
2, 3) are dependent.

Therefore, using the above equipment model equation, the process state T can be estimated by the following equation.

T1 = F11 (P ', F'), T2 = F32
(F ', N'), T3 = F11 (F21 (N ', F'),
F '), T4 = F11 (P', F31 (T ', N')), T
5 = F11 (F21 (N ', F'), F31 (T ', N')) where T ', P', F ', N'are observed signals of T, P, F, N (measured values ) And Ti is the process state estimated for T (i = 1, ..., 5).

It is not necessary to calculate dependent process states. For example, F11 (F21 (N ', F31 (T',
N ')) and F') are F11 (F21 (N ', F'), F31
Since it has a dependency relationship with (T ', N')), calculation is unnecessary.

Similarly, the process states of P, F and N can be estimated as follows.

P1 = F12 (T ', F'), P2 = F21
(N ', F'), P3 = F12 (F32 (F ', N'),
F '), P4 = F12 (T', F23 (P ', N')), P
5 = F12 (F32 (F ', N'), F23 (P ', N')) F1 = F13 (P ', T'), F2 = F23 (P ',
N '), F3 = F31 (T', N '), F4 = F13
(P ', F32 (F', N ')), F5 = F13 (F21
(N ', F'), T '), F6 = F13 (F21 (N',
F '), F32 (F', N ')) N1 = F22 (P', F '), N2 = F33 (T',
F '), N3 = F22 (F12 (T', F '), F'), N
4 = F22 (P ', F31 (T', N ')), N5 = F22
(F12 (T ', F'), F31 (T ', N') where Pi, Fi, Ni are process states (i = 1, ..., 5) estimated for P, F, N, respectively. .

Since there are a plurality of estimated values for each of T, P, F, and N as described above, when the observed signal becomes abnormal due to a sensor failure or the like in the estimated process state, the observed signal and equipment There is no coincidence with the value estimated by the model except coincidence. Therefore,
If any one of the estimated process states matches the observed signal, the observed signal can be considered normal.

The third step 103 is the processing of the observation signal diagnostic section 6, and the judgment is made by fuzzy inference.
The procedure of the judgment is shown below.

First, it is determined whether the process state is normal by using the process state (hereinafter referred to as the primary evaluation state) estimated by using one equipment model formula. Here, the state that "the first-order evaluation state is equal to the observed signal" is defined by the membership function, and it is considered that the grades determined by the membership function match.

For example, the membership function in the state of "T1 is equal to T '" is defined as shown in FIG. If the deviation between the estimated value T1 and the actually measured value is +5, the evaluation value T1 is regarded to match the observed signal T'at the clade 0.5.

Next, it is determined whether or not the process state (hereinafter, referred to as an Nth order evaluation state) estimated by using N (N is 2 or more) device model formulas is normal. Here, N
The fuzzy inference is performed based on the following logic, including the determination result of the evaluation state less than Nth order used in the next evaluation state.

Logic: "If all evaluation states less than the Nth order used in the Nth evaluation state are equal to the observation signal and the Nth evaluation state is equal to the observation signal, the Nth evaluation state matches the observation signal." Here, the state that "the Nth evaluation state is equal to the observed signal" is defined in the same manner as the membership function that represents the state that "the first evaluation state is equal to the observed signal". That is,
For example, the secondary evaluation state T2 is equal to the observed signal T'in the grade of 0.6, and the primary evaluation state F21 used in T2 is used.
When (N ', F') (= P2) is equal to the observed signal P'with a grade of 0.7, the MIN values of those grades are taken, and the evaluation value T2 becomes the observed signal T'with a grade of 0.6. Consider to be consistent.

Next, as the verification of the observation signal, it is determined whether the observation signal is normal or abnormal by fuzzy inference based on the determination results of all the evaluation states and the following logic.

Logic: "If any of the evaluation signals is equal to the observed signal, the observed signal is normal." That is, for example, T1 matches the observed signal T'with a grade of 0.5, and T2 is 0. Observation signal T ′ with a grade of 6
, T3 matches the observed signal T'at a grade of 0.4, T4 matches the observed signal T'at a grade of 0.7, and T5 matches the observed signal T'at a grade of 0.45. If so, the MAX values of those grades are taken, and the observed signal T ′ is assumed to be normal at grade 0.7.

Finally, the fourth step 104 is the processing of the status display section, which presents the diagnosis result of the sensor in the observation signal diagnosis section 6 to the operator together with the normal or abnormal grade.

By performing the above processing, in the case of a steady operation state, a sensor failure in consideration of the ambiguity included in the threshold value for the abnormality determination and the ambiguity included in the plant observation signal as noise using the equipment model. Can be diagnosed and the degree of possibility of sensor failure can be presented to the operator together.

FIG. 4 shows the configuration of the second embodiment according to the present invention. The plant operation support device 10 of this embodiment is
Since it has almost the same configuration as the plant operation support system 1 of the first embodiment, the same reference numerals are given to the same components and the description thereof will be omitted. As a point different from the first embodiment, the plant operation support apparatus 10 is an apparatus for inputting an observation signal determined to be normal by the observation signal diagnosis unit 6 and for determining the degree of change in the characteristics described in the equipment model by fuzzy inference. The characteristic diagnosis unit 11 and the state display unit 12 that displays the determination result of the device characteristic diagnosis unit 11 together with the output of the observation signal diagnosis unit
It has the point of having and.

The device characteristic diagnosing unit 11 evaluates the observation signals determined to be normal by the observation signal diagnosing unit 6 with a device model representing the relationship between these observation signals, and at the same time identifies the devices described in the device model by fuzzy inference. It is determined whether the characteristics have changed. That is, when the device characteristics shown in the device model used for state estimation change,
An error occurs between the process state estimated using the device model and the observed signal determined to be normal, and it is possible to identify which device characteristic has changed due to this error.

The processing of the device characteristic diagnosis section 11 will be described below.

If the characteristics of the equipment described in the equipment model formula have not changed, the relation of the equipment model formula evaluated using a normal observation signal should be satisfied. Therefore, the equipment model formula evaluated using the observation signal determined to be normal is
If the relationship is not satisfied, it can be considered that the characteristics of the equipment described in the equipment model formula are changing.

The fact that the characteristic of the device has changed means that the characteristic change of the device is determined by fuzzy inference based on the following logic, using the diagnosis result performed by the observation signal diagnosis section 6.

Logic: "When all the observation signals used in a certain equipment model formula are normal and the relation of the equipment model formula is not satisfied, the characteristics of the equipment described in the equipment model formula change. Here, the state of "not satisfying the relationship of the device model formula" is also defined by the membership function. For example, the membership function in the state of "not satisfying the relation of the equipment model formula T = F11 (P, F)" is defined as shown in FIG. 5, and the observation signal T'is normal in the grade of 0.4, The signal P'is normal in the grade of 0.6, the observed signal F'is normal in the grade of 0.8, and the equipment model formula T = F11
If (P, F) is grade 0.3 and does not satisfy the relationship,
Taking the MIN values of those grades, it can be considered that the characteristics of the equipment described in F11 are changing in the grade of 0.3.

On the other hand, the fact that the characteristics of the device have not changed is determined by fuzzy inference based on the following logic.

Logic: "When all the observation signals used in a certain equipment model formula are normal and the relation of the equipment model formula is satisfied, the characteristics of the equipment described in the equipment model formula are It has not changed. "Here, the state of" satisfying the relationship of the device model formula "is also defined by the membership function. That is, for example, the membership function in the state that "the device model equation T = F11 (P, F) is satisfied" is defined as shown in FIG. Normal, observation signal P'normal with a grade of 0.6, observation signal F'normal with a grade of 0.8, device model formula T =
If F11 (P, F) satisfies the relationship at grade 0.7, take the MIN value of those grades and consider that the characteristics of the equipment described in F11 are unchanged at the grade of 0.4. .

As described above, according to this embodiment, in the steady operation state, the soundness of the sensor is diagnosed by using the equipment model, and whether the characteristics of the equipment used in the equipment model are changed or not. It is possible to diagnose whether or not it is possible to present to the operator the degree of possibility of failure due to the change.

FIG. 7 shows the configuration of the third embodiment of the present invention. The plant operation support device 20 of the third embodiment
In addition to the configuration of the plant operation support device 10 according to the second embodiment, a normal signal identifying section 21 is provided. The normal signal identifying unit 21 determines that the observed signal is abnormal when the observed signal diagnostic unit 6 determines that the observed signal is abnormal and the device characteristic diagnostic unit 11 determines that the device characteristic has not changed. On the other hand, the degree of normality is identified by fuzzy reasoning. In addition, the status display unit 22 of the present embodiment
Is configured to display the observation signal, the determination result of the observation signal diagnosis unit 6, the determination result of the device characteristic diagnosis unit 11, and the identification value of the normal signal identification unit 21.

The processing of the normal signal identifying section 9 will be described below. The identification by the normal signal identifying unit 9 is performed by fuzzy inference based on the following logic.

Logic: "If there is no change in the characteristics of the equipment described in a certain equipment model equation and the observed signal used in the functional form of that equipment model equation is normal, it is estimated by that equipment model equation That is, the process state is correct. ”That is, for example, the characteristics of the equipment described in the equipment model formula T = F11 (P, F) are unchanged in the grade of 0.6, and the observation signal P ′ is in the grade of 0.5. If normal and the observed signal F'is normal with a grade of 0.8, the MIN values of those grades are taken and the process state T evaluated by F11 (P ', F') is correct with a grade of 0.5. Can be regarded as

By displaying this determination result on the state display unit 22, the normal value can be presented to the operator together with the grade of the possibility of normality of the observed signal determined to be abnormal, and the plant state can be grasped and grasped. This can be dealt with easily.

FIG. 8 shows the configuration of the fourth embodiment of the present invention in which a system exhibiting chaotic behavior is the object of diagnosis. The plant operation support device 30 of the present embodiment stores an output of a plant instrument, an input processing unit 31 that outputs an observation signal, and a characteristic amount of chaos such as the Lyapunov exponent of the observation signal during normal operation. Characteristic storage unit 32 and characteristic storage unit 3
Precision diagnosis unit for determining whether or not the plant state has changed from the plant state during normal operation by comparing the characteristic amount of chaos stored in 2 with the characteristic amount of chaos of the observed signal. 33 and a status display unit 34 that outputs the determination result of the precise diagnosis unit 33.

The processing of the precision diagnosis section 34 will be described in more detail below. Here, the unique feature of chaos becomes very clear by the time evolution of the system from two slightly different initial conditions. In this case, a non-chaotic system only leads to predictable errors over time. On the other hand, in a chaotic system, the error increases exponentially with time, and the state of the system becomes essentially unknown in a short time. This phenomenon, which occurs only when the system has nonlinearity, is called initial condition sensitivity, and the Lyapunov exponent λ is used to express the degree of this initial state sensitivity.
There is. This Lyapunov exponent λ is the sampling time τ
When N pieces of time series data (X1, X2, ..., XN) are obtained, the following is obtained.

[0061]

[Equation 1] Further, there is a q-th order Lyapunov exponent λq, which represents the degree of fluctuation of the Lyapunov exponent, and is calculated from the following equation.

[0062]

[Equation 2] By storing λ and λq in the steady state in the characteristic storage unit 4 in advance and comparing λ and λq obtained from the observed signal, it is possible to diagnose a change in the plant state.

As described above, according to the present embodiment, it becomes possible to monitor the state change of a system having a strong non-linearity which exhibits chaotic behavior or chaotic behavior which cannot be handled by the conventional linear method. .

FIG. 9 shows the configuration of the fifth embodiment of the present invention for diagnosing a plant in a transient state. The plant operation support device 40 of the present embodiment includes an input processing unit 41 that inputs an output of a plant instrument and outputs an observation signal, a characteristic storage unit 42 that stores a qualitative model that represents a dynamic characteristic of the observation signal, and an input. Observation signal from the processing unit 41, characteristic storage unit 4
2. A qualitative model is input from 2, and the behavior of the observed signal estimated by the qualitative model is compared with the actual observed signal to determine whether or not the observed signal is normal. The status display section 44 displays the judgment result of 43.

In the qualitative model of the characteristic storage unit 42, the factors that change the observed signal are described. This change factor is
Mutual interference between processes directly affects each other and some affects through a control system, and the factors that affect each case are described below.

In the transient diagnosis section 43, the soundness of the sensor is calculated by creating a dynamic propagation path between the observation signals based on the qualitative model stored in the characteristic storage section 42 and comparing it with the actual behavior. To diagnose.

On the other hand, as a method of describing the qualitative mutual relation of observed signals, a method of describing the changing direction of each observed signal by paying attention to an event in which a transient state occurs has been conventionally considered. In this case, there is an advantage that it is possible to identify events that have occurred due to changes in observed signals, but it is difficult to handle all possible events, and the description of observed signal changes must be redundant and the database must be used. Maintenance becomes difficult due to the huge number of items.

According to this embodiment, it is not possible to directly identify the occurrence event, but it is also possible to create a propagation path for changes in the observed signal, and a database is created without considering the overall spread. Therefore, the database can be easily maintained, and the failure of the sensor can be diagnosed during the transition by comparing it with the propagation path.

FIG. 10 shows the configuration of a sixth embodiment of the present invention for automatically determining the plant state and diagnosing the soundness of the observation signal of the dynamic or static characteristic according to the plant state. There is. The plant operation support device 50 according to the present embodiment inputs an output of a plant instrument and outputs an observation signal, and an observation signal of the input processing unit 51, inputs a observation signal of the plant to a steady state, a starting operation state, A plant state determination unit 52 that determines which one of transient states such as an output increase operating state, a device model of static characteristics of an observed signal, a qualitative model of dynamic characteristics, and a characteristic quantity indicating chaos of an observed signal. And a state estimation unit 54 that estimates a process state from an observation signal using an equipment model having a static characteristic when the plant state determination unit 52 determines that the plant is in a steady state, The observation signal diagnostic unit 55 that determines the degree of normality of the observation signal by comparing the process state estimated by the state estimation unit 54 with the observation signal, and the observation signal diagnostic unit 55 determines that the observation signal is normal. In the case where the device model is determined to be normal by the device characteristic diagnosing unit 56 and the device characteristic diagnosing unit 56, the observation is input using the observed signal. A normal signal identifying unit 57 that calculates a normal value for the observed signal determined to be abnormal by the signal diagnostic unit 55.
And a characteristic amount of chaos stored in the characteristic storage unit 53 when the observed signal exhibits chaotic behavior in a steady state,
When the precise diagnosis unit 58 that detects the change in the plant state by comparing the characteristic amounts of chaos of the observation signals and the plant state determination unit 52 determines that the plant is in the transient state,
A qualitative model indicating the dynamic characteristics of the observed signal is read from the characteristic storage unit 53, and the behavior of the observed signal estimated from the qualitative model is compared with the behavior of the actual observed signal to determine whether the observed signal is normal or not. From the transient diagnosis unit 59, the observation signal and the observation signal diagnosis unit 55, the device characteristic diagnosis unit 56, the normal signal identification unit 57, the precise diagnosis unit 58, and the status display unit 60 that displays the output of the transient diagnosis unit 59. Become.

The processing in the plant state judging section 52 of this embodiment will be described below.

The plant state judging section 52 is the input processing section 5
The rate of change of the observation signal input to 1 is monitored, and if the rate of change is within a certain value, it is determined that the plant state is in a steady state, otherwise the plant state is in a transient state. And Further, the observation signal representative of the operation mode among the observation signals is monitored, and it is determined whether the operation mode of the plant corresponds to any one of the starting operation state, the output increasing operation state, and the rated operation state. For example, when the observation signal representative of the operation mode is the plant output, when the plant output is 30% or less, the startup operation state, when 30% or more and less than 100%, the output operation state, and when 100%, the rating It is determined that the vehicle is operating.

According to the present embodiment, by the operation of the plant state judging section 52, the soundness of the sensor and the equipment is diagnosed according to the plant state, the operation mode, and the monitored object, and the observation signal of the abnormality is judged. It is possible to identify a normal value and monitor a change in plant state, and it is possible to automatically provide accurate plant information to an operator.

[0073]

As described above, according to the present invention,
It is possible to diagnose the health of sensors and devices by performing fuzzy inference using a quantitative static device model, and to identify normal values when a sensor is diagnosed as a fault, and In consideration of the ambiguity included in the observed signal as noise and the ambiguity of the threshold value for determining normality or abnormality, the possibility of normality or abnormality can be provided to the operator as a grade. In addition, transiently, a qualitative model showing the dynamic relationship of observed signals is used to diagnose changes in the plant state, and if the system to be monitored exhibits chaos or chaotic behavior, chaos By monitoring the characteristic quantities, it is possible to diagnose a system with a strong nonlinearity that could not be handled by the conventional linear method. Since accurate plant information can be automatically provided to the operator, it is possible to improve the safety of the plant while supporting the operator's situation judgment.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a plant operation support device according to a first embodiment of the present invention.

FIG. 2 is a flowchart showing a processing flow of the first embodiment of the present invention.

FIG. 3 is a graph showing a membership function that determines the degree to which the estimated value from the equipment model matches the observed signal.

FIG. 4 is a diagram showing a configuration of a plant operation support device according to a second embodiment of the present invention.

FIG. 5 is a graph showing a membership function for determining the degree of change of the device model.

FIG. 6 is a graph showing a membership function in which an observed signal satisfies a device model and determines a degree.

FIG. 7 is a diagram showing a configuration of a plant operation support device of a third embodiment of the present invention.

FIG. 8 is a diagram showing the configuration of a plant operation support system according to a fourth embodiment of the present invention.

FIG. 9 is a diagram showing a configuration of a plant operation support system according to a fifth embodiment of the present invention.

FIG. 10 is a diagram showing the configuration of a plant operation support system according to a sixth embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Plant operation support device 3 Input processing unit 4 Characteristic storage unit 5 State estimation unit 6 Observation signal diagnosis unit 7 State display unit 10 Plant operation support device 11 Equipment characteristic diagnosis unit 12 Status display unit 20 Plant operation support device 21 Normal signal identification unit 22 status display unit 30 plant operation support device 31 input processing unit 32 characteristic storage unit 33 precision diagnosis unit 34 state display unit 40 plant operation support device 41 input processing unit 42 characteristic storage unit 43 transient diagnosis unit 44 state display unit 50 plant operation Support device 51 Input processing unit 52 Plant state determination unit 53 Characteristic storage unit 54 State estimation unit 55 Observed signal diagnostic unit 56 Equipment characteristic diagnostic unit 57 Normal signal identification unit 58 Precision diagnostic unit 59 Transient diagnostic unit 60 Status display unit

Claims (7)

[Claims] 1. An input processing unit for inputting an output of an instrument of a plant and outputting an observation signal, a characteristic storage unit for storing a device model that quantitatively simulates a static characteristic of a component device of the plant, and the characteristic. A state estimation unit that estimates the process state of the plant from the observation signal of the input processing unit using the device model of the storage unit, compares the process state and the observation signal estimated by the state estimation unit, and fuzzy inference A plant operation support device, comprising: an observation signal diagnostic unit that determines the normality of the observation signal according to, and a status display unit that displays the observation signal and the determination result of the observation signal diagnostic unit. . 2. A device characteristic diagnosis unit that inputs an observation signal determined to be normal by the observation signal diagnosis unit to the device model and determines the degree of change in the characteristic of the device described in the device model by fuzzy inference. The state display unit is configured to display the observation signal, the determination result of the observation signal diagnostic unit, and the determination result of the device characteristic diagnostic unit. The plant operation support device described. 3. An observation signal is determined to be abnormal by the observation signal diagnosis unit in the input / output relationship of a predetermined component device,
And, when it is determined that the device characteristic has not changed by the device characteristic diagnosis unit, a normal signal identifying unit that identifies a normal value of the observed signal determined to be abnormal by the device model of the constituent device by fuzzy inference 3. The plant operation support device according to claim 2, wherein the state display unit is configured to display a normal value of the observation signal. 4. The characteristic storage unit stores a characteristic amount of chaos including a Lyapunov exponent during normal operation of an observation signal showing chaotic behavior, and the characteristic of the chaos stored in the characteristic storage unit. By comparing the characteristic amount and the characteristic amount of chaos of the observed signal, a precise diagnosis unit for determining whether or not the plant state has changed from the plant state during normal operation, the state display unit is The plant operation support device according to claim 1, wherein the plant operation support device is configured to display a determination result of the precise diagnosis unit. 5. The characteristic storage unit stores a qualitative model representing a dynamic characteristic of an observed signal, and the observed signal behavior and the actual observed signal behavior estimated by the qualitative model of the characteristic storage unit. And a transition time diagnosis unit that determines whether or not the observation signal is normal, and the state display unit is configured to display a determination result on the transition time diagnosis unit. The plant operation support device according to claim 1. 6. An input processing unit for inputting an output of an instrument of a plant and outputting an observation signal, a characteristic storage unit for storing a device model of a static characteristic of the observation signal and a qualitative model of a dynamic characteristic, and the input. A plant state determination unit that determines whether the plant is in a steady state or a transient state based on the observation signal of the processing unit, and when the plant state determination unit determines that the steady state,
A state estimation unit that estimates a process state of a plant from the observation signal using the equipment model, compares the process state and the observation signal estimated by the state estimation unit, and determines whether the observation signal is normal by fuzzy inference. An observation signal diagnostic unit that determines the degree, when the plant state determination unit determines that the transient state,
A transient diagnostic unit that determines whether or not the observed signal in a transient state is normal by comparing the observed signal behavior estimated by the qualitative model of the characteristic storage unit with the actual observed signal behavior, and the observed signal A plant operation support device comprising: a status display unit that displays the determination results of the observation signal diagnosis unit and the transient diagnosis unit. 7. The characteristic storage unit stores a characteristic amount of chaos of an observation signal including a Lyapunov exponent indicating a chaotic behavior in each operation mode including a startup operation state, an output increase operation state, and a rated operation state. Based on the determination result of the plant state determination unit, the characteristic amount of chaos corresponding to the operation mode and the characteristic amount of chaos of the observation signal are compared, and the norm in the corresponding operation mode is obtained. The plant operation according to claim 6, further comprising a precision diagnosis unit that determines a change from the plant state, wherein the state display unit is configured to display a determination result of the precision diagnosis unit. Support device.