JPH07234300A - Device for determining emergency operation in plant management and control system - Google Patents

Abstract

PURPOSE:To effectively determine an emergency operation by autonomously dealing with an emergency in a plant. CONSTITUTION:Equipment for which a suppressing operation is to be performed is determined by a performance detecting part 50 or a suppressed equipment determining part 53, and an effective controlled variable is determined by a first future value anticipating part 54 or a controlled variable determining part 59. In this case, the first future value anticipating part 54 obtains an anticipated value to deal with any unprecedented abnormal event using a neural network, and a second future value anticipating part 55 obtains an effective anticipated value using an approximate function to correspond to the features of input data. The anticipated values are corrected by an anticipated value correcting part 56 and fed to the controlled variable determining part 59. Using the anticipated values and a function of the influence of operations, the controlled variable determining part 59 determines the controlled variable based on a target value for a suppressing operation.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a plant management operation system for managing the operation of a plant such as a nuclear reactor plant, and more particularly to an emergency operation for determining an emergency operation corresponding to an abnormal situation occurring during plant operation. The present invention relates to an operation determining device.

[0002]

2. Description of the Related Art For example, a nuclear reactor plant such as a nuclear power plant belongs to a so-called enormous technology, and therefore, extremely high technical reliability is required for its development and practical application, and especially in its operation management. A high degree of safety is required. For example, if any abnormality occurs in the plant during operation of the nuclear reactor plant, it is necessary to first identify the cause and perform an operation to remove it. However, if the plant falls into an emergency state before the first cause is identified and there is no room to identify the cause, it is necessary to perform an operation for alleviating the observed event for the time being, that is, an emergency operation. For that purpose, it is necessary to promptly determine whether or not the plant operation is to be tripped or whether or not there is any operation for ensuring safety without tripping.

Conventionally, an emergency operation determination system capable of autonomously dealing with an emergency situation of a plant without making such a judgment manually has been proposed by the same inventor as the inventor of the present application, for example, in Japanese Patent Laid-Open No. 5-249287. Is proposed as. This emergency operation decision system has a knowledge base in which plant operation targets are hierarchically subdivided into a network, and the achievement level and urgency level are calculated based on this, which is effective for shifting to a more normal state. The target device for the abnormal condition suppression operation was specified. On the other hand, the temporal change of the abnormal plant variable from an appropriate point in the past to the present is approximated by a specific fixed function, for example, a quadratic function, and then,
Predict the future value of the plant variable by extrapolating this function in the future direction, and the predicted value, and the operation impact degree function that indicates the elapsed time of the plant measurement data when the suppression operation device is operated by a unit amount. Was used to determine the operation amount of the specified suppression operation target device so that the future value of the plant measurement data falls within a predetermined limit range.

[0004]

Since the conventional emergency operation determining system is configured as described above, it is possible to determine the operation equipment to be suppressed and the operation amount thereof regardless of human judgment in the emergency of the plant. You can However, when a complete emergency operation is performed when the plant is abnormal, there are the following problems.

(1) In the above-mentioned conventional technique, the future value of the plant variable is approximated and predicted by a specific fixed function, for example, a quadratic function. Therefore, when an unexperienced abnormal event is encountered online. It becomes difficult to predict,
When the input data that does not fit the fixed function is generated, the error of the predicted value may become extremely large, and when the manipulated variable is determined by this, there may be cases where complete emergency operations cannot be performed. . Therefore, it has been necessary to accurately calculate the predicted value in any abnormal situation.

(2) In the above-mentioned conventional technique, the predicted value of the plant variable is used only for determining the manipulated variable of the suppression control device only at that time. Therefore, when the predicted value has an error, the manipulated variable is calculated. There is a possibility that an error will occur in each determination of, and it will take time to achieve the optimum value of the operation target when the operation device to be suppressed is operated with the operation amount, and the optimum value cannot be achieved. There was a possibility. Therefore, it is necessary to increase the accuracy of the predicted value every time.

(3) In the above-mentioned conventional technique, the operation influence function used when determining the abnormal state suppression operation amount changes non-linearly depending on the plant state etc., but since this characteristic is not taken into consideration, There is a possibility that an error may occur in the determined operation amount.

(4) Since the conventional technique described above determines the operation amount using only the predicted value and the operation influence degree function,
It is not clear whether the manipulated variable determined by those values is appropriate or not, and the judgment criteria that determine at what point and to what extent the plant variable that is currently abnormal with this manipulated variable should be suppressed. In some cases, the appropriate amount of operation could not be determined.

(5) In the above-mentioned conventional technique, it is possible to obtain the allowable range of the abnormal state suppressing operation amount, but since the algorithm for obtaining the optimum value is not clear, for example,
In some cases, it was not possible to find a value that minimizes the risk of exceeding the limit value. Therefore, it may take time to obtain the optimum value when the suppression operation device is operated with the operation amount thus obtained.

(6) In the above-mentioned conventional technique, when the result of the abnormal state suppressing operation is poor, the cause cannot be diagnosed and appropriate countermeasures may not be taken thereafter.

(7) In the above-mentioned conventional technique, it may not be possible to determine a future countermeasure policy on the basis of the diagnosis result of the abnormal state suppressing operation failure, and appropriate countermeasure measures may not be taken thereafter. .

The inventions of claims 1 to 3 were made in order to solve the above problems, and it is predicted that any emergency situation of the plant can be dealt with when determining the operation amount of the emergency operation. An object of the present invention is to obtain an emergency operation determining device in a plant management operation system that can obtain a value with high accuracy and therefore can perform a complete emergency operation.

According to a fourth aspect of the present invention, it is possible to obtain an emergency operation determining device in a plant management operating system that can improve the accuracy even when there is an error in a predicted value and can quickly achieve an operation target. With the goal.

A fifth aspect of the present invention is a plant management operation capable of accurately generating an operation influence degree function that changes non-linearly depending on a plant state or the like, and reducing an error in an operation amount determined by the function. The purpose is to obtain an emergency operation determining device in the system.

According to the sixth aspect of the present invention, it is possible to determine at what point in time the plant variable that is currently abnormal should be suppressed and to what extent, and therefore, it is possible to appropriately determine the manipulated variable. The purpose is to obtain an emergency operation determining device in.

According to the invention of claim 7, it is possible to obtain a value that minimizes the risk of deviating from the limit value of the manipulated variable, and to quickly obtain the optimum value during the emergency operation. The purpose is to obtain an operation determining device.

It is an object of the present invention to provide an emergency operation determination device in a plant management operation system, which can diagnose the cause of abnormal results when abnormal condition suppression operations are defective.

According to the invention of claim 9, a future management policy can be decided based on a diagnosis result of an abnormal state suppressing operation failure cause, and thereby an appropriate future management measure can be taken. The purpose is to obtain an emergency operation determining device in.

[0019]

An emergency operation determining device in a plant management operating system according to the invention of claim 1 is
Equipment that is a target of abnormal state suppression operation that is effective to shift the operation target of the plant to a more normal state by hierarchically subdividing the operation target and obtaining the degree of achievement and urgency based on a knowledge base networked together. In addition to specifying the future value of the plant variable using a neural net and the operation impact degree function, the operation amount is determined so that the future value of the plant measurement data falls within a predetermined limit range. It is composed.

The emergency operation determining device in the plant management operation system according to the invention of claim 2 hierarchically subdivides the operation target of the plant and obtains the degree of achievement and the degree of urgency based on a knowledge base networked with each other. ,
In addition to identifying the device that is the target of the abnormal state suppression operation that is effective for shifting to a more normal state, use the value predicted using the approximation function determined based on the characteristics of the plant measurement data and the operation impact degree function. The operation amount is determined so that the future value of the plant measurement data falls within a predetermined limit range.

The emergency operation determining device in the plant management operation system according to the invention of claim 3 hierarchically subdivides the operation target of the plant and obtains the degree of achievement and the degree of urgency based on a knowledge base networked with each other. ,
In addition to identifying the equipment that is the target of the abnormal state suppression operation that is effective for shifting to a more normal state, the future value of the plant variable was determined based on the value predicted using a neural network and the characteristics of the plant measurement data. A value selected according to the situation using one of the values predicted using the approximation function, and the operation impact degree function is used to determine the operation amount so that the future value of the plant measurement data falls within the specified limit range. It is configured in.

In the emergency operation determining device in the plant management operating system according to the invention of claim 4, the predictive value correcting means connected to the future value predicting means uses the predictive value of the plant variable and the estimated value after the lapse of time. It is configured so that the predicted values of all plant measurement data that are abnormal due to the error characteristics with the measured values of plant variables are corrected, and the corrected predicted values are supplied to the manipulated variable determining means. is there.

The emergency operation determining device in the plant management operating system according to the fifth aspect of the present invention is a plant state based on the suppression operating device and the plant measurement data determined by the suppression operating device determining means by the influence degree function generating means. , The time-dependent change of the operation, and the operation influence degree function most suited to the current situation based on the simulated data stored in the database according to the information such as the operation direction.

In the emergency operation determining device in the plant management operation system according to the sixth aspect of the present invention, the operation target value calculating means should suppress the plant variable which is currently abnormal to any operation amount at any time. It is configured such that an abnormal state suppression operation target value defined by the suppression time point and the allowable upper limit value and lower limit value corresponding thereto is calculated and supplied to the operation amount determination means.

In the emergency operation determining device in the plant management operating system according to the seventh aspect of the present invention, the operation amount determining means determines the predicted value of the plant variable and the plant measurement data when the suppression operating device is operated by a unit amount. Suppression operation specified so that the risk that the future value of the plant measurement data will deviate from the predetermined limit range is minimized by solving the simultaneous linear equations by using the operation influence degree function indicating the change amount over time It is configured to determine the operation amount of the target device.

The emergency operation determination device in the plant management operation system according to the invention of claim 8 executes the operation when the operation amount is determined by the operation amount determination means in the defect cause diagnosis section. Detects whether the result is good, if the detection result is bad, diagnose the cause,
It is configured to be classified into an abnormal operation device, a defective input data, or a defective operation amount determination calculation.

The emergency operation determining device in the plant management operating system according to the ninth aspect of the present invention is based on the failure cause classified by the failure cause diagnosing section when the execution result is defective by the response policy determining section. It is configured to eliminate the cause of the defect by determining the countermeasure policy of (1) and controlling each part based on the determination result.

[0028]

The emergency operation determining device according to the invention of claim 1
Based on various plant measurement data periodically obtained for the plant operation by the achievement detection means, for each operation target located in the higher level in the knowledge base, the achievement level indicating the degree of normality of the state is obtained, and Whether or not the value of the achievement calculated by the achievement detecting means is within the predetermined range is detected by the urgency detection, and if it is within the predetermined range, it is determined that the emergency operation is possible although it is in an abnormal state. Then, the urgency is obtained by fuzzy inference with the achievement level, the rate of change of the achievement level indicating the change of the achievement level for each cycle, the importance degree for achieving the highest driving goal, and the sensitivity to disturbance as parameters. Next, in the operation target determining means, the urgency determined by the urgency detecting means is compared with the urgency of the driving target in which the other abnormality has occurred, and the operation in which the abnormality suppression is given the highest priority By determining a target and performing a normality check on the operating targets that belong to the lower levels of the operating target determined by the operating target determining unit based on the knowledge base by the suppression operation device determining unit, a more normal state is obtained. Identify the device that is the target of the abnormal state suppression operation that is effective for the transition. On the other hand, in the neural network of the future value prediction means, time series data obtained by sampling plant measurement data of the plant variable from the present time to the past at appropriate intervals is input to the input layer, and this output is output via the intermediate layer to the output layer. To output to the output layer the time-series data that predicts future measurement data. At this time, the coupling degree between the input layer and the output layer can be changed appropriately in the intermediate layer so that all abnormal plant measurement data can be accurately predicted. In addition, an operation impact function indicating the change over time of the plant measurement data when the suppression operation device determined by the suppression operation device determination means is operated by a unit amount by the impact function function generation means is generated, and the future value is generated. The value predicted by the predicting means and the operation influence degree function generated by the influence degree function generating means are supplied to the operation amount determining means. Thus, the operation amount determining unit determines the operation amount of the suppression operation target device specified by the suppression operation device determining unit such that the future value of the plant measurement data falls within the predetermined limit range.

In the emergency operation determining device according to the second aspect of the present invention, based on various plant measurement data periodically obtained for the plant operation by the achievement degree detecting means, each operation target positioned at a higher level in the knowledge base. In addition, the achievement level indicating the degree of normality of the state is obtained, and the urgency level detection unit detects whether the value of the achievement level obtained by the achievement level detection unit is within a predetermined range, If it is within the range, it is judged that emergency operation is possible even though it is in an abnormal state, and the degree of achievement, the rate of change in the degree of achievement indicating the change in the degree of achievement for each cycle, and the importance for achieving the highest operating goal , And the degree of urgency is obtained by fuzzy inference using parameters such as sensitivity to disturbance. Next, the operation target determining means compares the urgent level obtained by the urgent level detecting means with the urgent level of the driving target in which another abnormality has occurred, and the operating target for which the abnormality suppression is given the highest priority. Then, based on the knowledge base by the suppression operation device determination means, a normality check is sequentially performed on the operation targets that are subordinate to the operation target determined by the operation target determination means, and a more normal state is entered. The device that is the target of the abnormal state suppression operation that is effective for the purpose is specified. On the other hand, the feature of the time series data obtained by sampling the measured data of the plant variable from the present time to the proper interval is extracted by the future value prediction means, and the type of the approximation function is determined based on this feature data, and the approximation error is determined. Predict all abnormal plant measurement data by using the function that defines the parameters of the approximation function so that the square of is minimized. In addition, the influence function generation means generates an operation influence function indicating a time lapse of the variation amount of the plant measurement data when the suppression operation device determined by the suppression operation device determination means is operated by a unit amount, and these future functions are generated. The value predicted by the value predicting means and the operation influence degree function generated by the influence degree function generating means are supplied to the operation amount determining means. As a result, the operation amount determination means determines the operation amount of the suppression operation target device specified by the suppression operation device determination means such that the future value of the plant measurement data falls within the predetermined limit range.

In the emergency operation determining device according to the third aspect of the present invention, based on various plant measurement data periodically obtained for the plant operation by the achievement degree detecting means, each operation target positioned at a higher level in the knowledge base In addition, the achievement level indicating the degree of normality of the state is obtained, and the urgency level detection unit detects whether the value of the achievement level obtained by the achievement level detection unit is within a predetermined range, If it is within the range, it is judged that emergency operation is possible even though it is in an abnormal state, and the degree of achievement, the rate of change in the degree of achievement indicating the change in the degree of achievement for each cycle, and the importance for achieving the highest operating goal , And the degree of urgency is obtained by fuzzy inference using parameters such as sensitivity to disturbance. Next, the operation target determining means compares and evaluates the urgency obtained by the urgency detection means and the urgency of the other operation target in which an abnormality has occurred, and the operation in which the abnormality suppression should be given the highest priority. By determining a target and performing a normality check on the operating targets that belong to the lower levels of the operating target determined by the operating target determining unit based on the knowledge base by the suppression operation device determining unit, a more normal state is obtained. Identify the device that is the target of the abnormal state suppression operation that is effective for the transition. On the other hand, the plant measurement data is sequentially supplied to the first future value prediction means and the second future value prediction means, and the predicted value is calculated. That is, in the neural network of the first future value prediction means, time series data obtained by sampling plant measurement data of the plant variable from the present time to the past at appropriate intervals is input to the input layer, and this output is passed through the intermediate layer. The time series data, which is supplied to the output layer and predicts future measurement data, is output from the output layer. At this time, the coupling degree between the input layer and the output layer can be changed appropriately in the intermediate layer so that all abnormal plant measurement data can be accurately predicted. Further, the second future value prediction means extracts the characteristic of the time series data obtained by sampling the measured data of the plant variable from the present time to the past at appropriate intervals, and determines the type of the approximation function based on the characteristic data. Then, all the abnormal plant measurement data are predicted using the function that defines the parameters of the approximation function so that the square of the approximation error is minimized. These predicted values are supplied to the manipulated variable determining means. In addition, the influence function generation means generates an operation influence function indicating the elapsed time of the variation amount of the plant measurement data when the suppression operation device determined by the suppression operation device determination means is operated by a unit amount, and this is generated. It is supplied to the manipulated variable determining means. Thereby, the manipulated variable determining means selects either the value predicted by the first future value predicting means or the value predicted by the second future value predicting means, and the future value and the influence degree function are generated. The operation amount of the suppression operation target device specified by the suppression operation device determination means is determined so that the future value of the plant measurement data falls within a predetermined limit range using the operation influence degree function from the unit.

In the emergency operation determining device according to the fourth aspect of the present invention, the predictive value correcting means connected to the future value predicting means,
The predicted value of all plant measurement data that is abnormal due to the error characteristic between the predicted value of the plant variable and the measured value of the plant variable after a lapse of time is calculated, and the corrected predicted value Is supplied to the manipulated variable determining means. As a result, even if there is an error in the predicted plant variable, the error is corrected and supplied to the operation amount determining means, and the operation amount determining means can determine the operation amount with high accuracy.

In the emergency operation determining device according to the invention of claim 5, the influence function generating means determines the suppression operation equipment determined by the suppression operation equipment determining means, the plant state based on the plant measurement data, and the time change of the operation. And the operation influence degree function that best matches the current situation is generated based on the simulated data stored in the database according to the information such as the operation direction. Thus, the operation amount determining means is supplied with the operation influence degree function adapted to the non-linearly changing plant state, and can determine the operation amount suitable for the current plant state.

In the emergency operation determining device according to the sixth aspect of the present invention, the operation target value calculating means determines at what time and for which operation amount the plant variable that is currently abnormal should be suppressed, at the time of the suppression time. An abnormal state suppression operation target value specified by the allowable upper limit value and the lower limit value corresponding thereto is calculated and supplied to the operation amount determining means. As a result, the operation amount determination means can determine an appropriate operation amount based on the abnormal state suppression operation target value calculated by the operation target value calculation means.

In the emergency operation determining apparatus according to the invention of claim 7, the operation amount determining means determines the predicted value of the plant variable and the change over time of the change amount of the plant measurement data when the suppression operation device is operated by a unit amount. The operation amount of the suppression operation target device specified so that the risk that the future value of the plant measurement data will deviate from the predetermined limit range is minimized by solving the simultaneous linear equations using the operation influence function shown decide.

In the emergency operation determining device according to the invention of claim 8, when the defect cause diagnosing section performs the operation with the operation amount determined by the operation amount determining means, whether or not the execution result is good. Is detected, and if the detection result is defective, the cause is diagnosed and classified into an abnormal operation device, a defective input data, or a defective operation amount determination calculation. This allows
By removing the cause of the defect, the subsequent emergency operation can be improved.

In the emergency operation determining device according to the invention of claim 9, when the execution result of the emergency operation is defective, the response policy determining unit determines the future response policy based on the defect cause classified by the defect cause diagnosing unit. By determining and controlling each part based on this determination result, the cause of the defect is removed,
The first aid operation afterwards can be improved.

[0037]

【Example】

Example 1. An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a functional block diagram showing an embodiment of an emergency operation device in a plant management operation system of the present invention, and FIG. 2 is a block diagram showing a plant management operation system to which the device is applied. In FIG. 2, reference numeral 10 denotes an emergency operation device that determines an emergency operation when a plant abnormality is detected, and stores a central processing unit 11 that performs various kinds of calculation and judgment processing, and information necessary for the processing. It has an external storage device (storage means) 12, a main storage device 13 that stores programs of the central processing unit 11 and the like, and a plant data input interface 14 that supplies data from the plant to the central processing unit 11. Reference numeral 15 denotes a plant data collection device that collects various plant measurement data representing the state of the plant. For example, the state of the device and the flow rate of the fluid through a measuring instrument installed in the plant operation device and valves. The process data representing the state is detected and supplied to the emergency operation device 10. Reference numeral 16 denotes a plant operation device that performs actual plant operation management and includes an operation control panel and the like.

In this embodiment, the external storage device 12 has
An operation target function hierarchical knowledge base (hereinafter referred to as a knowledge base) 17 as shown in FIG. 3 is stored. The knowledge base 17 is a knowledge base created based on data relating to the function and structure of the plant in order to grasp the situation of the plant, and as shown in FIG. 3, an operation target level 21, a plant level 22, a control system. Level 23,
And the device level 24 has a structure in which four layers are layered and networked with each other. In other words, each level consists of one or more nodes that represent more specific plant operating goals and means for achieving those goals, with higher nodes representing more comprehensive content and lower nodes being more comprehensive. It shows the specific content. Each node selectively has the following information necessary for operation determination according to its type. (1) Node information: information indicating its own node, information indicating an immediately upper node, and information for calculating an immediately lower node. (2) Node status information: information indicating whether the node is normal or abnormal, and includes information indicating the degree of urgency in the case of abnormality, information indicating soundness, and the like. (3) Equivalent operation amount calculation information; information for calculating an operation amount required to change or maintain node information. (4) System identification information; this is information indicating which system the node belongs to. (5) Device operation information: information required to operate an actual device.

More specifically, each node of the knowledge base 17 in this embodiment is divided into the following five types. The node 31 is a driving target node, and here, for example, the maintenance of the output driving state is the driving target. In addition, the nodes 32-1 and 32-2 are nodes indicating elements for achieving a higher driving goal, and are called state evaluation nodes. Here, for example, there are a heat sink control node, a furnace heat output control node, and the like. A soundness evaluation node is further positioned below the state evaluation node. For example, below the heat sink control node 32-1 which is one of the state evaluation nodes, the secondary Na flow rate securing node 33-
1 to main cooler outlet Na temperature control node 33-3 and the like. Below the soundness evaluation node are relay nodes 34-1 to 34-3.
34-3 is located. Further, a plurality of suppression operation nodes are provided below the relay node.
For example, at the lower level of the Na temperature suppression operation node 34-2, 1
A main cooler equivalent suppression operation device node 35-1 and the like are arranged. A device for actually performing an operation is arranged below the suppression operation node. For example, the inlet vane operating device 36-1 and the like are arranged below the 2A main cooler equivalent suppression operating device node 35-2. In addition, another state evaluation node, for example, the furnace heat output control node 32
-2 also has the same hierarchical structure. Note that Na mentioned in the above description means liquid sodium used as a coolant for taking out the heat energy generated by nuclear fission from the core to the outside, and the main cooler means an intermediate heat exchanger. A heat exchanger and the like for cooling the heat of the core sent via the air is shown.

The central processing unit 11 in this embodiment has a functional configuration as shown in FIG. 1, for example. In FIG. 1, reference numeral 50 is an achievement level detection unit (achievement level detection means) that obtains the achievement level indicating the degree of normality of the state for each operation target located in the higher order in the knowledge base 17, and the plant data input interface 14 It is an arithmetic processing unit that receives the plant measurement data from the plant data collection device 15 via the, and obtains the achievement level by a predetermined arithmetic processing. For example,
Using the plant evaluation index I represented by a function with plant data as an independent variable, a higher-level node is selected that is more comprehensive and can evaluate a wider range of the plant, and this is normalized from the variable x. The value of the state evaluation function f (x) as shown in FIG. 7 is obtained, and the degree of achievement is obtained based on this state evaluation function f (x).

Further, reference numeral 51 determines whether or not the value of the achievement level obtained by the achievement level detection unit 50 is within a predetermined range, and if it is within the predetermined range, an emergency operation is carried out although it is in an abnormal state. Is an urgent level detection unit (urgency level detection unit) that determines the urgent level of the driving target by a predetermined calculation process. In the present embodiment, the achievement level obtained by the achievement level detection unit 50, the achievement level change rate indicating the change in the achievement level for each cycle, the importance level for achieving the highest driving goal, the sensitivity to disturbance, and the like. The degree of urgency is obtained by fuzzy inference using the parameter as a parameter. The achievement rate of change indicates how much the achievement value in one cycle changed in the next cycle. If the change is negative, it affects the urgency. . The importance level indicates the importance level for achieving the driving goal, and the higher the node, the greater the value. Sensitivity is a measure of how sensitively a response to a disturbance is, and the higher the value, the more urgent the influence is.

Reference numeral 52 denotes a driving target determining unit (driving target determining means) which determines the driving target for which abnormality suppression is to be given the highest priority by comparing and weighing the urgency levels of the respective driving targets obtained by the urgency detecting unit 51. ). In other words, the driving target determining unit 52 detects the achievement levels of the respective driving targets sequentially by the achievement level detecting unit 50, and when the urgency level detecting unit 51 sequentially detects the urgency level for an abnormality, When there are abnormalities in a plurality of operation targets, the comparison processing unit determines the operation target for which abnormality suppression is to be given the highest priority from those. 53
Is an abnormality effective for shifting to a more normal state by sequentially checking the normality of the target level belonging to the lower level of the driving target determined by the driving target determining unit 52 based on the knowledge base 17. It is a suppression operation device determination unit (suppression operation device determination means) that specifies a device that is a target of the state suppression operation.

Reference numerals 54 and 55 denote suppression operation device determining section 5.
It is a future value prediction unit (future value prediction means) that predicts the future value of the abnormal plant variable corresponding to the combination of the highest priority order of the suppression operation devices determined in 3. A first future value prediction unit (first future value prediction unit) 54 using a net and a second future value prediction unit (second future value prediction unit) 5 to which an approximate function is applied
5 and 5. The first future value prediction unit 54, as shown in FIG. 4, time-series data x ₁ , x obtained by sampling the measured data of the plant variable from the present time to the past at appropriate intervals.
₂ ,. ． ． x _m as an input layer In and time series data y ₁ , y ₂ ,. ． ． The output layer Out that outputs y _n , and the input layer In and the output layer O
ut and the intermediate layer M which is coupled with variable coupling strength
It has an id. In this neural network, for various abnormal events, the actually generated measurement data is used as a teacher signal, and the coupling strength is learned so that the predicted data matches the teacher signal as much as possible. In this case, various learning methods are conceivable, but in the present embodiment, as an example, from the output layer In of the neural network to the input layer Out.
The backpropagation method is adopted in which the elements of each layer, that is, the coupling strength between the neurons are changed in the opposite direction, and the learning is repeated by repeating this.

The second future value prediction unit 55 extracts the characteristics of the time series data obtained by sampling the plant measurement data received through the plant data input interface 14 at appropriate intervals, and determines the type of the approximation function. This is an arithmetic processing unit that predicts all abnormal measurement data using the function that defines the parameters of the approximation function so that the square of the approximation error is minimized. That is, the characteristics of the measured data of plant variables for various abnormal events, for example,
By analyzing the increase / decrease of the entire trend, the presence / absence of vibration, the increase / decrease of amplitude, etc., pattern classification is performed, and a database is established that clarifies the type of the approximation function with the time that best matches this as an independent variable. There are the following three patterns as the approximate function type determination knowledge adopted in this embodiment. (1) If the measured data increases monotonically and gently, a quadratic function is adopted. (2) If the measured data decreases monotonically and rapidly, the exponential function is adopted. (3) If the measured data oscillates with its amplitude rapidly increasing, the product of the exponential function and the sine function is adopted.

Reference numeral 56 designates the first future value predictor 54 and the second future value predictor 54.
Is a prediction value correction unit (prediction value correction means) that corrects the prediction value from the future value prediction unit 55, and is an arithmetic processing unit that corrects the prediction value according to those error characteristics obtained in advance. In the present embodiment, for example, when a positive or negative error continuously occurs in the predicted value, it is determined as a systematic error, the predicted value is corrected in the direction of reducing the error, and the predicted value is positive. Alternatively, if the negative error occurs irregularly, it is determined to be a random error and is supplied to the next stage without correction.

Reference numeral 57 indicates an influence degree function generating section (a degree influence function generating section for generating an operation influence degree function indicating the passage of time of the variation amount of the plant measurement data when the suppression operation apparatus determined by the suppression operation apparatus determining section 53 is operated by a unit amount. In the present embodiment, it is a function generation unit that generates an operation influence function that best matches the current situation according to the plant state that changes non-linearly by a plant simulator or the like. For example, the plant simulator creates a database of suppression operation equipment, plant status, time change of operation, and operation direction, and when the data of the current plant status is supplied, select the item that matches the current situation. ,
The operation influence function is calculated based on them.

Reference numeral 58 denotes an operation target value calculation for calculating an abnormal state suppression operation target value which defines at what time point and to what extent the abnormal plant variable should be suppressed by the suppression time point and the allowable upper and lower limit values corresponding thereto. Part (operation target value calculation means), for example, in the present embodiment, the suppression operation is determined based on the suppression operation determination time, the operation execution time, the operation delay time, the predicted value of the abnormal plant variable, and the operation influence degree function. It is an arithmetic processing unit that calculates a target value.

Reference numeral 59 denotes an operation amount determination unit (operation amount determination means) for determining the operation amount of the specified suppression operation device, and the first future value prediction unit 54 received via the predicted value correction unit 56.
Alternatively, the predicted value from the second future value prediction unit 55, the suppression operation influence function from the influence function generation unit 57, and the abnormal state suppression operation target value from the operation target value calculation unit 58 are received, and these are used to perform simultaneous 1 It is an arithmetic processing unit that determines the manipulated variable by the following equation.

Reference numeral 60 denotes a defect cause diagnosis unit (defect cause diagnosis unit) for diagnosing, by a predetermined calculation, whether or not the result is satisfactorily performed when an emergency operation is executed with the operation amount of the operation amount determination unit 59. It is a means) and diagnoses whether it is good or not based on the plant measurement data after operation and the predicted value, and if the result is bad, diagnoses whether the cause of the failure lies in the operating device, input data or manipulated variable. It is the diagnostic department.

Reference numeral 61 is a countermeasure policy determining section (corresponding policy determining means) which determines a countermeasure policy thereafter based on the result of diagnosis by the failure cause diagnosis section 60. For example, when there is a cause in the operating device, knowledge is given. When information such as the inoperability of the operating device is registered in the data, and when there is a cause in the input data, information such as inability to measure the detecting device or inoperability of the operating device is registered, and when there is a cause in the operation amount. Is a control processing unit that causes the predicted value correction unit 56 and the like to perform corresponding processing such as supplying data that further supplements the correction amount.

Next, the operation will be described. FIG. 5 is a flowchart for explaining the operation when the above system is applied to, for example, "Joyo" which is an experimental reactor of a fast breeder reactor.
In the figure, first, in step S101, the central processing unit 11 activates the plant data collection system 15 at a constant cycle. As a result, the plant data collection system 15 collects necessary data of the plant each time it is activated and sends it to the central processing unit 11 via the plant data input interface 14. Upon receiving this, the central processing unit 11 performs the processing shown in FIG.
The achievement degree AF of the state evaluation node is calculated (step S10).
2). Here, as an example, a case will be described in which the achievement level is obtained when the heat sink control node 32-1 in FIG. 3 is used as a state evaluation node.

In FIG. 6, when the plant data is acquired (step S201), first, the achievement degree detecting unit 50 obtains the plant evaluation index I based on the following equation (1) (step S202).

I = F (Ei−Eo) (1) where F represents the secondary system Na flow rate, and Ei and Eo
Represent the main chiller inlet Na enthalpy and the main cooler outlet Na enthalpy, respectively. The plant evaluation index I obtained from this represents the amount of heat removed by the so-called main cooler here. In this way, as the plant evaluation index I, a plant data is generally expressed by a function having an independent variable, and a higher-rank node that is more comprehensive and can evaluate a wider range of the plant is selected.

Next, in the achievement level detecting section 50, the following (2)
The plant evaluation index I is normalized based on the formula and the independent variable x
Is calculated (step S203).

X = (I-In) / In (2) Here, In represents a normal value of the plant evaluation index.

Next, the achievement degree detecting unit 50 obtains the value of the state evaluation function f (x) as shown in FIG. 7 for the independent variable x thus obtained (step S20).
4). This state evaluation function f (x) is a dead zone lower limit value x ₂
From 0 to the dead zone upper limit value x ₃ , the value is "0", the value is "-1" for the trip lower limit value x ₁ or less, and "1" for the trip upper limit value x ₄ or more. Take

Next, the achievement level detecting section 50 obtains the achievement level AF from the value f (x) of the state evaluation function obtained in this way based on the following equation (3) (step S20).
5).

AF = 1− | f (x) | ... (3)

Therefore, the state evaluation function f (x) is "±
In the case of 1 ", the achievement level is 0. On the contrary, when f (x) is 0, the achievement level AF takes the value of" 1 ". Dan's urgency detector 5
1 is supplied.

Next, the urgency detecting section 51 determines what kind of emergency operation should be performed next based on the value of the achievement degree AF thus obtained. That is, when the achievement degree AF is "1" (step S103; Y), it is determined that the operation target is 100% achieved and is in a normal state (step S104), and no particular emergency operation is determined. The processing of the cycle ends.

On the other hand, when the achievement degree AF is "0" (step S105; Y), it is determined that no emergency operation is possible, and it is determined that the emergency stop should be performed (step S106). Then, the timing of the emergency stop is determined (step S107), and the plant operation system 16
Instructing to perform an emergency stop at the determined timing.

Further, when the achievement degree AF has a value between "0" and "1", the urgency level detecting section 51 determines that some emergency operation is possible and the urgency level EF is determined.
Is calculated (step S108). As will be described later, the calculation of the urgency level EF can be performed by fuzzy reasoning, for example, in the present embodiment. The obtained urgency levels are sequentially supplied to the operation target determination unit 52.

If another abnormal node is generated at the stage of obtaining the urgency EF (step S109;
Y), the operation target determining unit 52 compares the urgency with each other (step S110), and when the urgency of the new abnormal node is higher (step S110; N), the priority is given to the abnormal node. To migrate. On the other hand, when the abnormal node has a higher urgency (step S110;
Y), the abnormal node is determined as a target for the emergency operation (step S111). Information on the determined abnormal node is further supplied to the suppression operation device determination unit 53.

When the abnormal node is judged to be abnormal in this way, the suppression operation device determining section 53 then suppresses the degree of the abnormal state and shifts to a more normal state. A node for performing an operation (hereinafter referred to as an abnormal state suppression operation node) is determined using the knowledge base 17 according to the procedure (FIG. 8) described later (step S11).
2). Then, the operation amount for the determined abnormal state suppression operation node is determined by the procedure (FIG. 9) described later (step S113). In this way
When an abnormality occurs, it is determined whether an operation to suppress the abnormal state is possible or an emergency stop is required, and if an emergency operation is possible,
Specifically, the target to be operated is specified and the operation amount is determined.

Next, in step S108, the urgency level E
The procedure for obtaining F will be described below. As described above, here, so-called fuzzy inference is used to execute the urgency EF.
Ask for. At this time, the achievement degree AF, the achievement degree change rate AF ′, and the sensitivity RF are used as parameters for the fuzzy function.
To use. Then, the degree of urgency is obtained based on fuzzy inference using a rule group as shown in the following Tables 1 to 4.

[0066]

[Table 1]

[0067]

[Table 2]

[0068]

[Table 3]

[0069]

[Table 4]

As described above, in fuzzy inference, various fuzzy variables are set, but each variable has a plurality of areas,
For example, regarding the achievement degree AF, it is represented in three areas of high, middle, and low, and the urgency EF corresponding thereto is also divided into three areas of low, middle, and high. Similarly, the degree of achievement change AF '
Also, the importance level IF and the sensitivity level RF are represented for each of a plurality of regions. A so-called membership function is defined as follows for each of these areas. That is, the membership function summarized for each variable used in this inference is as shown in the following expression (4).

H _i (y) = Σ _j f _ij (x _i ) · g _ij (y) (j = 1 to 3) (4) where x _i is the current value of the i-th variable. , Y indicates the degree of urgency. Further, h _i is a membership function indicating the degree of influence of the i-th variable on the urgency, and f i
_ij is a membership function indicating the contents of the j-th area of the i-th antecedent variable, and g _ij is j of the i-th antecedent variable.
Membership function indicating the contents of the second domain. Here, the antecedent part variable is, for example, the achievement degree AF in Table 1.
And the consequent part variable is the urgency EF. Note that all variables and function values are normalized so as to take values from "0" to "1". However, when it takes a negative value,
It takes "1" from "-1".

Then, a membership function indicating the degree of influence of all variables on the urgency EF is obtained from the following equation (5).

H _t (y) = Σ _i W _i · h _i (y) (i = 1 to 4) (5) where ht is the degree of influence of all variables on the urgency. Is a member sheep function shown, and W _i is a weighting coefficient showing the degree of influence of the i-th variable on the urgency. Then, the estimated value y * of the urgency is obtained by the following equation (6), and this is adopted as the urgency EF.

Y * = ∫h _t (y) · ydy / ∫h _t (y) · dy ... (6)

Next, the procedure for determining the abnormal state suppression operation node in step S112 will be described with reference to FIG. First, the suppression operation device determination unit 53 starts by searching for a soundness evaluation node that is currently abnormal. The soundness evaluation node indicates, for example, the main cooler outlet Na temperature control node 33-3 in FIG. 3 as described above, and by searching for this, the target achievement level of the system can be evaluated and the abnormal system can be identified. Will also be possible. In this process, first,
The node is checked (step S301), and if not abnormal (step S302; N), the next node is checked and if abnormal (step S302; N).
Y), it is determined whether the node is an AND node or an OR node.

Here, the AND node means that all lower nodes belonging to the node must be healthy in order to secure the soundness of the node.
In order to ensure the soundness of the node, the OR node is a substitute for another node if any one of the nodes belonging to the subordinate node is sound. However, it must be available to be an alternative.

As a result, if the node is an OR node (step S303; N), flags are set for all available lower nodes and a priority is determined using a certain rule (step S304). . On the other hand, if it is an AND node (step S30
3; Y). One lower node is extracted (step S30
5) Check the node (step S30)
6). As a result, if the lower node is abnormal (step S307; Y), the flag is set to the lower node (step S308), and if not abnormal (step S307; N), the flag is not set. . And
The processes from step S305 to step S309 are performed until the check of all the subordinate nodes to which the accident belongs is completed.

The above-mentioned processing is sequentially performed for all the soundness evaluation nodes that are currently abnormal until the abnormal state suppression operation node is reached. Then, the suppression operation node with the flag set becomes the target of the abnormal state suppression operation.

Next, the determination of the operation amount for the abnormal state suppressing operation in step S113 will be described with reference to FIG. First, among the abnormal state suppression operation nodes flagged in step S112, the combination of the highest priority is extracted (step S401). Then, the future value of the abnormal plant variable corresponding to this combination is predicted by the first future value prediction unit 54 and the second future value prediction unit 55 (step S402). At this time, the features of the two prediction units are taken into consideration and used properly depending on the situation.

(A) Prediction Using Neural Network Time series data obtained by sampling measurement data of the plant variable from the present time to the past at appropriate intervals is used as an input layer, and time series data predicted of future measurement data is used as an output layer, Create the neural net shown in FIG. 4 with an intermediate layer. Next, for various abnormal events, the actually generated measurement data is used as a teacher signal, and the coupling strength is learned so that the predicted data matches the teacher signal as much as possible. In this case, the backpropagation method is adopted in which learning is performed by changing the coupling strength between the neurons in the opposite direction from the output layer to the input layer of the neural network and repeating the learning.
A future value is predicted online as follows using the above neural network. (1) Input past time series data. (2) Output data is calculated from input data using the learned binding strength.

When the future value is predicted by the first future value prediction unit 54, it has the following advantages. (1) It is possible to predict even if an abnormal event that has not been experienced is encountered while online. (2) Since it has a learning function, it is possible to deal with a wider variety of abnormal events as it gains experience. On the other hand, there are the following problems. (1) It takes time to create a program and learn the bond strength in advance. (2) Only the sample time point learned in advance can be predicted,
Other time points need to be interpolated or extrapolated from the values at the sample time points. Therefore, the first future value predicting unit 54 is rarely used near the time of starting the device, and the second future value predicting unit 55 to which the following approximation function is applied is applied near the time of starting the device to It is advisable to switch from the second future value prediction unit 55 when the future value prediction unit 54 has sufficiently learned.

(B) Prediction Using Approximation Function In the second future value prediction unit 55, first, the characteristics of measured data of plant variables for various abnormal events, such as increase / decrease of the entire trend, presence / absence of vibration, and amplitude Patterns are classified by analyzing increase / decrease, etc., the type of approximation function with the best matching time as an independent variable is clarified, the knowledge base is constructed, and the quadratic function, exponential function and exponential function and sine function described above are constructed. Is adopted according to the characteristics of the input measurement data, and the future value is predicted as follows. (1) Input past time series data. (2) The feature of the input data is extracted, and the type of the approximate function to be adopted is determined using the approximate function type determination knowledge. (3) The approximation error at the time of inputting the time-series data is squared, and the parameters of the approximation function are obtained so that the function obtained by summing all of them is minimized. An example of approximation with a quadratic equation is shown below.

Time is x-axis (horizontal axis), plant variable value is y
For the axis (vertical axis), y is approximated by the following equation (7) with y as a function of x.

Y = f (x) = ax ² + bx + c (7) Here, the parameters a, b, and c are determined so that the following expression (8) is minimized.

Σ _i (f (x _i ) −y _i ) ² (i = 1 to N) (8) where i is the sample time point number, N is the number of samples, and x
_i is the time at the time of sampling, and y _i is the input value of the plant variable at the time of sampling. Specifically, the following differential equations (9) to (12) obtained by partially differentiating the parameters a, b, and c and setting the differential value to zero are solved.

CN + b [x] + a [x ² ] = [y] (9)

C [x] + b [x ² ] + a [x ³ ] = [xy] ... (10)

C [x ² ] + b [x ³ ] + a [x ⁴ ] = [x ² y] (11) where [f (x)] = Σ _i f (x _i ) (i = 1 ~ N) ... (12)

The second future value prediction unit 55 has the following advantages as compared with the first future value prediction unit 54. (1) Prototype creation is easy, and stepwise development is possible based on which a knowledge base is gradually added. (2) Since an approximate function is used, any time can be predicted. On the other hand, there are the following problems. (1) In order to accurately predict various abnormal events, it is necessary to add a knowledge base. (2) When an abnormal event that has not been experienced is encountered while online, there is no guarantee that it can be accurately predicted. Therefore, when the apparatus is started, the knowledge base of the second future value prediction unit 55 is used in advance in a known range, and is switched according to the learning degree of the first future value prediction unit 54. After an unexperienced event occurs, the knowledge base of the second future value prediction unit 55 is upgraded and adopted based on the experience. By repeating these, it is possible to improve the prediction accuracy of the first future value prediction unit 54 and the second future value prediction unit 55.

Next, when the apparatus is started up and after an unexperienced event has occurred, the first future value prediction unit 54
Since an error has occurred in the second future value prediction unit 55, in that case, the predicted value correction unit 56 corrects the predicted value. That is, the error between the predicted value of the plant variable and the measured value of the plant variable after a lapse of time has already been calculated by the predicted value correction unit 56 (step S50).
In the case of 7), based on this, the predicted value obtained in step S402 is corrected as necessary according to the error characteristic as follows (step S403). (1) If a positive or negative error occurs continuously, it is determined that the error is a systematic error, and the predicted value is corrected so as to reduce the error. (2) If positive or negative errors occur irregularly, it is judged as a random error and no correction is made. The predicted value thus corrected is sequentially supplied to the manipulated variable determiner 59.

On the other hand, in the influence function generator 57,
An operation influence degree function that matches the current situation is determined (step S404). In this case, the plant state is obtained in advance by using a plant simulator or the like, and the database is created according to the following items. (1) Suppression operation device (2) Plant information obtained from output value, measured data such as steady / unsteady or normal / abnormal, strictly speaking, an infinite number of states are conceivable, but it is a major item because the data amount becomes excessive ,
For example, only output values are taken and expressed discretely. (3) Time change of operation of steps, lamps, etc. (4) Direction of operation such as pulling out / inserting of control rod and opening / closing of valve

Then, using the above database, the operation influence degree function is determined as follows. (1) Select items that currently match the situation. (2) The database is searched according to the selected item, and the operation influence degree function is obtained. At this time, if there is no function that exactly matches the item, the function corresponding to the item adjacent thereto is interpolated. For example, currently 9
When there is only a function corresponding to 100% and 90% output when the output is 6%, both functions are interpolated to obtain a function corresponding to 96% output.

Next, the operation target value calculation unit 58 determines the operation target value for suppressing an abnormal state, that is, at what point and at what point the plant variable that is currently abnormal should be suppressed. The value defined by is determined as an appropriate value according to the situation (step S405). In this case, if the target value is fixed as in the conventional case, the following problems occur.

1. If the interval from the present time to the suppression time is too short, the following problems will occur. (1) It is necessary to frequently perform the calculation for determining the suppression operation, which increases the computer load. (2) Due to the suppression operation determination time, the operation execution time, and the delay time until the operation affects the plant variables,
There may be no operation that achieves the target value. 2. If the interval from the current time point to the suppression time point is too long, the following problems occur. (1) The error between the predicted value and the calculated value of the plant variable becomes large. (2) In the case of the step operation, the effect of the operation on the plant variables decreases and saturates after a certain time elapses after the operation, so that the expected effect cannot be obtained. 3. As a value that regulates how much should be suppressed in this operation, it is desirable that the closer to the normal value the normal value can be returned quickly, but if the value too close to the normal value causes an overshoot, there is a problem. Occurs. on the other hand,
If the value is too far from the normal value, when the plant variable suddenly becomes abnormal, there is a risk that the suppression operation will not be in time and the allowable value will be exceeded.

Therefore, the operation target value calculation unit 58 determines the suppression operation target value based on the following information while considering the above restrictions. (1) Suppression operation determination time, operation execution time, operation delay time (2) Predicted value of abnormal plant variable (3) Degree of overshoot of operation impact degree function, degree of saturation after a certain period of time)

Next, the manipulated variable determiner 59 determines the present time t _i.
From the operation time t _max (step S405) to the next (1
A suppression operation amount that always satisfies the condition shown in the expression 3) is obtained (step S406).

X _imin ≦ x _i (t) + Σ _j Σ _k δ _{i, k} · U _j (t) · S _{j, k} (t) ≦ x _imax (j = 1 to jmax, k = 1 to kmax)・ ・ (13)

Here, i indicates a sound correct evaluation node that is currently abnormal, and j indicates a suppression operation node for suppressing them. Further, x _i (t) indicates the predicted value of the plant variable of the abnormal node i (steps S402 and S4).
03). X _imin is the present time t _{i of the} plant variable
Indicates the lower limit of the allowable range from the operation time t _max to x,
Similarly, _max indicates the upper limit value (step S405). Further, δ _{i, k} is 1 only when i = k and is 0 otherwise. S _{j, k}
(T) represents the operation influence degree function (step S404) given to the node k when the suppression operation node j is operated, and U
_j (t) represents the input function of the suppression operation node j (function indicating how many times the input is the operation influence function).

In the equation (13), everything except the input function is known, and an appropriate sample time (t ₁ , ..., T _max ) is selected from the present time to the suppression time point, and the number of unknowns at each time point is selected. Is j _max and the number of equations i _max is solved, and if a solution exists, the following (14)-
The solution of equation (15) is obtained.

U _jmin (t ₁ ) ≦ U _j (t ₁ ) ≦ U _jmax (t ₁ ) (14) ...

U _jmin (t _max ) ≦ U _j (t _max ) ≦ U _jmax (t _max ) ... (15)

The most desirable solution among the above is (13)
The one that minimizes the risk of violating the formula is given. The input function U _j is a value U _jmin (t1) that determines the lower limit of the solution, ...
, U _jmax (t _max ) has a maximum value of U _{jmin · max} , while U _jmax (t ₁ ), ..., U _jmax (t
_If the minimum value of ( _max ) is U _{jmax · min} , it can be obtained from the following equation (16).

U _j = (U _{jmin · max} + U _{jmax · min} ) /2.0 (16)

If there is an operation amount that satisfies the above conditions (step S407; Y), it is determined whether the operation amount can be operated (step S408). When the operation amount can be operated, the operation amount is adopted, and the operation device that cannot be used, the priority order of the operation device that can be used under the OR connection, the constraint required by the plant, etc. In consideration of the above, a specific operation method is determined based on the operating procedure and the like, and an instruction to execute the operation is issued to the plant operation system to be executed (step S409). On the other hand, when the operation amount satisfying the expression (13) cannot be found (step S407; N) and the operation of the operation amount is impossible (step S408; N), a combination of other suppression operation nodes is extracted. (Step S410), the processing from step S402 onward is performed.

Next, the defect cause diagnosing unit 60 judges the quality of the executed suppression operation as follows based on the following equations (17) and (18) (step S411).

| E (t) | = | x _{i · ins} (t) −x _{i · pre} (t) | ≦ ε (17)

| E (t) | = | x _{i · ins} (t) −x _{i · pre} (t) |> ε (18)

In the case of the expression (17), it is judged as good, and (1
In the case of 8), it is determined to be defective. Where x _{i · ins} (t)
Is the measured value of the plant variable of the abnormal node i when the suppression operation is executed, x _{i · pre} (t) is the predicted value of the same variable, and e (t)
Is the difference between the two, that is, the error in the measured value, and ε is the boundary value for determining whether it is good or bad.

When it is determined that the condition is good, it is determined whether or not the suppressing operation is necessary (step S413). If the plant has settled in the ideal state, the process ends. If not, the processes in and after step S402 are performed.

On the other hand, when it is determined that the result is a defect, the cause of the defect in the suppression operation result is diagnosed (step S414). This diagnosis method will be described with reference to FIG.

First, it is checked whether or not the determined operation is faithfully executed (step S501). When the determination operation is not faithfully executed (step S502; N)
Identifies the operating device in which the operating device is abnormal, for example, the valve is considered to be stuck open or the like. On the other hand, when the determination operation is faithfully executed (step S502; Y), the validity of the input data used to determine the operation amount is checked based on the signal validation result (step S504).

When the input data, that is, the plant measurement data is not valid (step S505; N), the abnormal input data is identified. On the other hand, when the input data is valid (step S505; Y), the defect of the manipulated variable determination calculation is identified as follows (step S50).
7). (1) It is checked whether the used operation influence degree function matches the current situation, and the validity of the function is judged. If they do not match, recalculate a valid function. (2) The error between the predicted value of the plant variable and the measured value of the plant variable after the elapse of time is calculated using the operation influence function that has been calculated again as necessary. This error is step S
In 403, it is used to correct the predicted value.

Further, the countermeasure policy determining section 61 determines a future countermeasure policy based on the diagnosis result of the cause of the defect as follows. (Step S415). (1) When the operating device is abnormal If the operating device has redundancy and an alternative device is available, use it. When the operation node cannot be used, it is determined that the operation node cannot be used. (2) When the input data is not valid If there is redundancy in the input data and alternative data can be used, use that. If the operation node cannot be used, it is determined that the operation node cannot be used. (3) When the operation amount determination calculation is defective, the operation node is determined to be usable, but the predicted value of the plant variable obtained by using the operation influence degree function recalculated as necessary and the time The error from the measured value of the plant variable (step S414) is used for the correction of the predicted value to be obtained (step S403).

Further, it is determined whether or not the operation node can be used (step S416). If the operation node can be used, the process of step S413 and the following steps is performed, and if the operation node is not available, the operation of step S410 and the following steps are performed.

In the above embodiments, the case where the present system is applied to the fast breeder reactor, for example, the experimental reactor "JOYO" has been described as an example, but the system of the present invention is not limited to this and other nuclear power plants are used. It may be applied to a power generation plant, a thermal power generation plant, a chemical plant, or the like. In these cases, the same operation can be performed by creating a knowledge base matching each plant.

Further, in the above embodiment, the first future value predicting section 5 by the neural network is used as the future value predicting means.
4 and the second future value prediction unit 55 based on the approximation function are appropriately switched and used according to the situation. In the present invention, either the first future value prediction unit 54 or the second future value prediction unit 55 is used. Either one may be used.

[0117]

As described above, according to the invention of claim 1,
Equipment that is a target of abnormal state suppression operation that is effective to shift the operation target of the plant to a more normal state by hierarchically subdividing the operation target and obtaining the degree of achievement and urgency based on a knowledge base networked together. In addition to specifying the future value of the plant variable using a neural net and the operation impact degree function, the operation amount is determined so that the future value of the plant measurement data falls within a predetermined limit range. Since it is configured, it is possible to obtain an accurate future value even if an unexperienced abnormal event is encountered, and to determine the operating device and the operation amount that can effectively suppress the abnormal state regardless of human judgment. is there.

According to the second aspect of the present invention, the operation target of the plant is subdivided hierarchically, and the achievement level and the urgency level are obtained based on the knowledge bases networked with each other, so as to shift to a more normal state. In addition to identifying the equipment that is the target of the abnormal state suppression operation that is effective for, the value predicted using the approximation function determined based on the characteristics of the plant measurement data and the future value of the plant measurement data using the operation impact degree function Since it is configured to determine the operation amount so that it falls within the predetermined limit range, even if an unexperienced event is encountered, an accurate future value is sought and the abnormal state is effectively suppressed regardless of human judgment. There is an effect that it is possible to determine the operation device and the operation amount that can be performed.

According to the third aspect of the present invention, the operation target of the plant is hierarchically subdivided, and the achievement level and the urgency level are obtained based on the knowledge bases networked with each other to shift to a more normal state. In addition to identifying the equipment that is the target of the effective abnormal state suppression operation, the future value of the plant variable is predicted using a neural network and the value predicted using the approximation function determined based on the characteristics of the plant measurement data. One of the values is selected according to the situation, and the operation amount is configured to determine the operation amount so that the future value of the plant measurement data falls within the predetermined limit range by using the operation influence function. There is an effect that the operation device and the operation amount capable of effectively suppressing the abnormal state can be determined by obtaining the value regardless of the judgment of the person.

According to the fourth aspect of the invention, the error characteristic between the predicted value of the plant variable and the measured value of the plant variable after a lapse of time is calculated by the predicted value correction means connected to the future value prediction means. There is an error in the predicted plant variables because the predicted values of all abnormal plant measurement data are calculated and the corrected predicted values are supplied to the manipulated variable determining means. Even in such a case, the error is corrected and supplied to the operation amount determining means, and the operation amount determining means can determine the operation amount with high accuracy.

According to the invention of claim 5, the influence function generating means determines the suppression operation device determined by the suppression operation device determination means, the plant state based on the plant measurement data, the time change of the operation, and the operation time. Since the operation influence function that best matches the current situation is generated based on the simulated data stored in the database according to the information such as the direction, the operation amount determining means is adapted to the non-linearly changing plant state. An operation influence function can be supplied, which has the effect of determining an operation amount suitable for the current plant state.

According to the sixth aspect of the present invention, the operation target value calculating means determines at what time and at what time the operation amount of the plant variable that is currently abnormal should be suppressed, corresponding to the suppression time and the operation amount. Since the abnormal state suppressing operation target value defined by the allowable upper limit value and the lower limit value is calculated and supplied to the operation amount determining means, the operation amount determining means causes the abnormality calculated by the operation target value calculating means. There is an effect that an appropriate operation amount can be determined based on the state suppression operation target value.

According to the seventh aspect of the present invention, the operation amount determining means has an operation influence indicating the predicted value of the plant variable and the change amount of the plant measurement data when the suppression operation device is operated by a unit amount. By determining the operation amount of the suppression operation target device specified so that the risk that the future value of the plant measurement data will deviate from the predetermined limit range is minimized by solving the simultaneous linear equations using the degree function and Since it is configured as described above, there is an effect that the emergency operation of the abnormal plant can be quickly performed.

According to the invention of claim 8, when the operation is executed with the operation amount determined by the operation amount determining means, the defect cause diagnosis unit detects whether or not the execution result is good. , If the detection result is bad, diagnose the cause,
Since it is configured to classify the operation device as abnormal, the input data is defective, or the manipulated variable determination calculation is defective, when the emergency operation is performed, the cause can be identified when the result is defective. Therefore, by removing the cause of the defect, it is possible to improve the subsequent emergency operation.

According to the ninth aspect of the present invention, in the case where the execution result is defective, the countermeasure policy determining unit determines a future countermeasure policy based on the defect causes classified by the defect cause diagnosing unit, and makes this determination. By controlling each part based on the result, the cause of the defect is eliminated, so that there is an effect that the emergency operation can be favorably performed after the emergency operation is performed.

[Brief description of drawings]

FIG. 1 is a functional block diagram showing an emergency operation determination device in a plant management operation system according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a plant management operation system to which the emergency operation determination device in the embodiment of FIG. 1 is applied.

FIG. 3 is an explanatory diagram showing an example of a driving target function layered knowledge base in the embodiment of FIG.

4 is a structural diagram showing a neural network applied to the embodiment of FIG. 1. FIG.

FIG. 5 is a flowchart showing the overall process in the embodiment of FIG.

FIG. 6 is a flowchart showing details of the processing content of step S102 in the processing of FIG.

FIG. 7 is an explanatory diagram showing a state evaluation function applied in the embodiment of FIG.

8 is a flowchart showing details of the processing content of step S112 in the processing of FIG.

9 is a flowchart showing details of the processing content of step S113 in the processing of FIG.

10 is a flowchart showing details of the processing content of step S414 in the processing of FIG.

[Explanation of symbols]

12 External Storage Device (Storage Means) 17 Operating Goal Function Hierarchical Knowledge Base (Knowledge Base) 50 Achievement Level Detection Unit (Achievement Level Detection Unit) 51 Urgency Level Detection Unit (Emergency Level Detection Unit) 52 Driving Target Determining Unit (Driving Target) Determining means) 53 Suppression operation equipment determining section (suppression operation equipment determining means) 54 First future value prediction section (future value prediction means, first future value prediction means) 55 Second future value prediction section (future value prediction) Means, second future value prediction means) 56 Prediction value correction unit (prediction value correction means) 57 Influence degree function generation unit (impact degree function generation means) 58 Operation target value calculation unit (operation target value calculation means) 59 Operation amount Determining part (manipulation amount determining means) 60 Defect cause diagnosing part (defect cause diagnosing means) 61 Corresponding policy deciding part (corresponding policy deciding means)

Claims (9)

[Claims] 1. Receiving plant measurement data representing the state of the plant, detecting whether or not an abnormality has occurred in the plant based on the plant measurement data, and determining an emergency operation when the abnormality has occurred. In the emergency operation determination device in the plant management operation system, storage means for hierarchically subdividing the operation target of the plant based on the function, structure, operation manual, etc. of the plant and storing a knowledge base networked with each other, Achievement degree detection means for obtaining an achievement degree indicating the degree of normality of the state for each operation target positioned at a higher position in the knowledge base based on various plant measurement data periodically obtained for plant operation, and It is detected whether the value of the achievement calculated by the achievement detecting means is within a predetermined range, and if it is within the predetermined range, an abnormal state is detected. Although it is determined that emergency operation is possible, the achievement level, the rate of change in the achievement level indicating the change in the achievement level for each cycle, the importance level for achieving the highest driving goal, and the sensitivity to disturbances, etc. Priority is given to the urgency detection means for determining the urgency by fuzzy reasoning as a parameter, and the urgency obtained by the urgency detection means and the urgency of the driving target in which another abnormality has occurred are weighed and given the highest priority. The operating target determining means for determining the operating target for which abnormal suppression is to be performed, and the normality check for the operating targets belonging to the subordinates of the operating target determined by the operating target determining means based on the knowledge base. By performing the suppression operation device determination means for identifying the device that is the target of the abnormal state suppression operation that is effective for shifting to a more normal state,
It is a future value prediction means that predicts all abnormal plant measurement data, and the input layer that inputs the time series data obtained by sampling the past plant measurement data from the current plant variable at appropriate intervals, and the future The future of predicting future values with a neural network that has an output layer that outputs time-series data that predicts measurement data, and an intermediate layer that connects the input layer and the output layer with a connection strength that can be changed appropriately A value predicting means, and an influence degree function generating means for generating an operation influence degree function indicating a time lapse of a variation amount of the plant measurement data when the suppression operation equipment determined by the suppression operation equipment determination means is operated by a unit amount. , The future value of the plant measurement data is determined by using the value predicted by the future value predicting means and the operation influence degree function generated by the influence degree function generating means. And an operation amount determining means for determining the operation amount of the suppression operation target device specified by the suppression operation device determining means so as to be within the limit range of the emergency operation determining device in the plant management operation system. . 2. Receiving plant measurement data representing the state of the plant, detecting whether or not an abnormality has occurred in the plant based on the plant measurement data, and determining the emergency operation when the abnormality has occurred. In the emergency operation determination device in the plant management operation system, storage means for hierarchically subdividing the operation target of the plant based on the function, structure, operation manual, etc. of the plant and storing a knowledge base networked with each other, Achievement degree detection means for obtaining an achievement degree indicating the degree of normality of the state for each operation target positioned at a higher position in the knowledge base based on various plant measurement data periodically obtained for plant operation, and It is detected whether the value of the achievement calculated by the achievement detecting means is within a predetermined range, and if it is within the predetermined range, an abnormal state is detected. Although it is determined that emergency operation is possible, the achievement level, the rate of change in the achievement level indicating the change in the achievement level for each cycle, the importance level for achieving the highest driving goal, and the sensitivity to disturbances, etc. Priority is given to the urgency detection means for obtaining the urgency by fuzzy reasoning as a parameter, and the urgency obtained by the urgency detection means and the urgency of the driving target in which another abnormality has occurred are weighed and given the highest priority. The operating target determining means for determining the operating target for which abnormal suppression is to be performed, and the normality check for the operating targets belonging to the subordinates of the operating target determined by the operating target determining means based on the knowledge base. By performing the suppression operation device determination means for identifying the device that is the target of the abnormal state suppression operation that is effective for shifting to a more normal state,
A future value predicting unit that predicts all abnormal plant measurement data, and extracts the characteristics of the time series data obtained by sampling the past measurement data of the plant variables at appropriate intervals, and extracting the characteristic data. A future value predicting unit that predicts a future value using the function that determines the type of the approximation function based on the above, and determines the parameter of the approximation function so that the square of the approximation error is minimized, and the suppression operation device determination. Predicted by the influence value function generating means for generating the operation influence degree function indicating the time lapse of the variation amount of the plant measurement data when the suppression operation device determined by the means is operated by a unit amount, and the future value prediction means. The suppression operation device is determined so that the future value of the plant measurement data falls within a predetermined limit range using the value and the operation influence function generated by the influence function generating means. Emergency operation determination device in the plant management control system, characterized in that it comprises a manipulated variable determining means for determining an operation amount of the specified inhibited the target device at stages. 3. Receiving the plant measurement data representing the state of the plant, detecting whether or not an abnormality has occurred in the plant based on the plant measurement data, and determining the emergency operation when the abnormality has occurred. In the emergency operation determination device in the plant management operation system, storage means for hierarchically subdividing the operation target of the plant based on the function, structure, operation manual, etc. of the plant and storing a knowledge base networked with each other, Achievement degree detection means for obtaining an achievement degree indicating the degree of normality of the state for each operation target positioned at a higher position in the knowledge base based on various plant measurement data periodically obtained for plant operation, and It is detected whether the value of the achievement calculated by the achievement detecting means is within a predetermined range, and if it is within the predetermined range, an abnormal state is detected. Although it is determined that emergency operation is possible, the achievement level, the rate of change in the achievement level indicating the change in the achievement level for each cycle, the importance level for achieving the highest driving goal, and the sensitivity to disturbances, etc. Priority is given to the urgency detection means for obtaining the urgency by fuzzy reasoning as a parameter, and the urgency obtained by the urgency detection means and the urgency of the driving target in which another abnormality has occurred are weighed and given the highest priority. The operating target determining means for determining the operating target for which abnormal suppression is to be performed, and the normality check for the operating targets belonging to the subordinates of the operating target determined by the operating target determining means based on the knowledge base. By performing the suppression operation device determination means for identifying the device that is the target of the abnormal state suppression operation that is effective for shifting to a more normal state,
An input layer that inputs time series data obtained by sampling past plant measurement data from the present time of plant variables at appropriate intervals, an output layer that outputs time series data that predicts future measurement data, and these input and output layers A first future value predicting means for predicting all abnormal plant measurement data by using a neural network having an intermediate layer connecting between and with an appropriately variable degree of connection; The characteristics of the time-series data obtained by sampling the past measurement data at appropriate intervals from are extracted, the type of the approximation function is determined based on the characteristics data, and the square of the approximation error is minimized. Second future value predicting means for predicting all abnormal plant measurement data using the function defining the parameter, and the suppression operation device determination The influence degree function generating means for generating an operation influence degree function indicating the time lapse of the variation amount of the plant measurement data when the suppression operation device determined by the means is operated by a unit amount, and the first future value predicting means. Either the predicted value or the value predicted by the second future value prediction means is selected according to the situation, and the predicted value and the operation influence function generated by the influence function generation means are used. And a manipulated variable determining means for determining the manipulated variable of the suppression operation target device specified by the suppression operation device determining means so that the future value of the plant measurement data falls within a predetermined limit range. Emergency operation determination device in plant management operation system. 4. The future value predicting means predicts the predicted values of all plant measurement data that are abnormal due to an error characteristic between a predicted value of a plant variable and a measured value of the plant variable after a lapse of time. 4. A predictive value correcting means for obtaining a corrected value is connected to each of the predictive value correcting means, and the predictive value corrected by the predictive value correcting means is supplied to the manipulated variable determining means. An emergency operation determining device in the plant management operation system according to claim 1. 5. The database according to information such as the suppression operation device determined by the suppression operation device determination means, the plant state based on the plant measurement data, the time change of the operation, and the operation direction, and the like. The emergency operation determining device in the plant management operating system according to any one of claims 1 to 3, wherein an operation impact function that best matches the current situation is generated based on the simulated data. . 6. An abnormal state suppression operation target that defines at what point in time a plant variable that is currently abnormal should be suppressed and to what manipulated variable it is defined by a suppression point and corresponding allowable upper and lower limit values. The emergency operation determination device in the plant management operation system according to any one of claims 1 to 3, further comprising operation target value calculation means for calculating a value. 7. The operation amount determining means uses the predicted values of the plant variables and an operation influence degree function indicating a time lapse of a change amount of the plant measurement data when the suppression operation device is operated by a unit amount, and the simultaneous operation is performed. By solving a linear equation, the operation amount of the specified suppression operation target device is determined so that the risk that the future value of the plant measurement data deviates from a predetermined limit range is minimized. An emergency operation determination device in the plant management operation system according to any one of claims 1 to 6. 8. When an operation is executed with the operation amount determined by the operation amount determining means, it is detected whether or not the execution result is good, and if the detection result is bad, the cause thereof is detected. The plant management according to any one of claims 1 to 7, further comprising: a failure cause diagnosing means for diagnosing and categorizing a malfunction of an operating device, a defect of input data, or a defect of calculation calculation of an operation amount. Emergency operation determination device in operating system. 9. When the execution result is defective, a countermeasure policy determining means for determining a future countermeasure policy based on the fault cause classified by the fault cause diagnosis unit, and controlling each unit based on the determination result. The emergency operation determination device in the plant management operation system according to claim 8, further comprising: