JPH02259803A - Plant controller - Google Patents

Abstract

PURPOSE:To perform the optimum control operation even when an unexpected state occurs by utilizing profound knowledge that is a base to decide the structure of a plant, the characteristic of equipment, experimental knowledge, and related regulations, etc. CONSTITUTION:A state monitoring function 3 recognizes the state of the plant 1, and sends change data to a counter plan setting function 7 when the plant falls in an unexpected abnormal state. The counter plan setting function 7 is comprised of a profound control knowledge base 10 which stores the profound knowledge that is the base to decide control operations of the structural model of the plant, the dynamic characteristic model of the equipment, the experimental knowledge, and the related regulations, etc., in a normal state, and an inference part 9, and predicts the future state of the plant by starting up a predictive simulation function 8, and generates a control target and the control operation required for recovery by utilizing the profound knowledge. In such a way, it is possible to cope with the state of the plant flexibly, and to safely and stably perform the optimum control operation.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to an automatic operation device for a plant, and particularly to a control expert system.

(Prior art) In recent years, with the development of machine technology, knowledge engineering has been introduced to automatic operation of thermal power plants, and an expert system for controlling thermal power plants has been proposed (Japanese Patent Publication No. 56-356).
Publication No. 1) has already been put into practical use. (Toshiba Shiv, L
+, Vol, 29. No. 10, P. 845~
850, 1974> A thermal power plant (hereinafter simply abbreviated as a plant) is
In addition to main equipment such as boilers, turbines, generators, transformers, and circuit breakers, it consists of a large number of auxiliary equipment. Therefore, in order to maintain the optimal operating conditions of these plants, each plant component must be adapted to the constantly changing operating conditions of the plant, based on its characteristics, operational standards, etc. For this reason, the optimal operation is determined in advance based on the characteristics of the plant component equipment, operational standards, etc., depending on the plant status and timing at that time. Specifications are determined and designed based on the characteristics of the structure and equipment, as well as empirical knowledge accumulated over many years and knowledge of related laws and regulations. The prior art will be explained below with reference to FIG. The thermal power plant control device has status monitoring function 3
and a control action function 4, which receives plant data from the thermal power plant 1, issues control commands in accordance with the plant data, and controls the plant 1 in an optimal state.

The condition monitoring function 3 takes in plant data from the plant 1, monitors the plant condition, detects change data, and transfers the data to the control action function 4. The control action function 4 is composed of an inference section 5 and a control knowledge base 6. The control knowledge base 6 stores control knowledge whose specifications are determined and designed based on knowledge of the above-mentioned plant structure, i-equipment characteristics, empirical knowledge, related regulations, and the like. The reasoning unit 5 takes in change data from the status monitoring function 3, recognizes the status of the plant 1, determines a knowledge command based on the control knowledge stored in advance in the control knowledge base 6, and outputs it to the plant 1. The state of the plant changes as external factors such as equipment operations or input commands change, but as mentioned above, the state monitoring function 3 monitors the plant state. The control action function 4 determines a control command based on the control knowledge stored in the control knowledge base 6 in advance, and repeats the output cycle. Therefore, such a power generation plant control expert system can perform optimal operation according to the ever-changing plant conditions. Furthermore, even if an abnormal situation occurs, as long as the operation is determined in accordance with the plant status at that time, optimal operations such as recovery or shutdown can be performed. Furthermore, if the operation has not been determined in advance in an abnormal situation, the control is suspended (locked) and left to the operator's discretion.

(Problems to be Solved by the Invention) In the conventional power plant control expert system described above, the operation is based on the structure of the plant and the characteristics of the WA device.

Limited to those determined in advance based on empirical knowledge and knowledge of related laws and regulations. Therefore, in the event of an unexpected situation, safety control measures such as stopping or suspending (locking) are adopted, which may not necessarily be the optimal control operation.

The present invention has been made in view of the above circumstances, and is based on plant structure 1, characteristics of equipment, and empirical knowledge.

Our goal is to provide plant control equipment that enables optimal control operations even in unforeseen situations by utilizing in-depth knowledge of related laws and regulations that form the basis for determining operations. .

[Structure of the Invention] (Means for Solving the Problem) In order to achieve the above object, the present invention includes a state monitoring function that detects the state of the plant, and a control operation that is performed based on experiential knowledge according to the plant state. In plant control equipment equipped with control action functions, we have deep knowledge of the predictive simulation function that predicts the future state of the plant, and the control objectives and control operations necessary for recovery in the event of an unexpected situation by inputting the current and future plant states. It has been configured to add a countermeasure planning function that generates based on.

(Function) When an unexpected abnormal condition occurs, change data is sent to the countermeasure planning function. Here, there is a deep control knowledge base that stores deep knowledge such as plant structural models, equipment dynamic characteristic models, empirical knowledge, and related laws and regulations on which control operations are determined. Therefore, the countermeasure planning function recognizes the state of the plant, activates the predictive simulation function to predict the future state of the plant, and generates control targets and fn scraping necessary for recovery using deep knowledge.

(Example) Examples will be described below with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of a plant control device according to the present invention. In FIG. 1, parts that are the same as those in FIG. 3 are given the same reference numerals and explanations are omitted.

7 is a countermeasure planning function, which is composed of an inference section 9 and a deep control knowledge base 10, where the deep control knowledge base 10 includes control information such as a structural model of a plant, a dynamic characteristic model of an Il, experiential knowledge, and related laws and regulations. Stores deep control knowledge that is the basis for determining operations. 8 is a predictive simulation function, which predicts the future state of the plant. The other configurations are the same as in FIG. 3.

In Fig. 1, the condition monitoring function 3 recognizes the condition of the plant, determines whether an abnormality in the plant condition is detected, and if an abnormality is detected, it is an unexpected state that causes an emergency stop or interruption. or whether it is a simple abnormal state for which countermeasures have been determined in advance. As a result, if the plant is in a normal or simply abnormal state, change data is sent to the control action function 4.

On the other hand, in the case of an unexpected abnormal state, change data is sent to the countermeasure planning function 7. The countermeasure planning function 7 includes a deep control knowledge base 10 that stores deep knowledge such as plant structural models, equipment dynamic characteristic models, empirical knowledge, and related laws and regulations that are the basis for determining control operations during normal times, and an inference unit. Consists of 9.

FIG. 2 is a flowchart showing the inference process, and the processing of the inference section 9 will be explained using this flowchart. In 821, the inference unit 9 identifies the location of the failure in the plant 1 based on the change data from the condition monitoring function 3.

In S22, based on the plant operating principles (principles that determine the priority of the plant operating status, such as ``achieving the operating schedule'' or ``maintaining the status quo''), the plant 1 is asked ``What state do you want it to be in?'' A request is generated.

In S23, in order to satisfy the generated request, an operation is identified based on deep knowledge of the connection relationship of devices, characteristics of the devices, and the like. In other words, since deep knowledge describes the connection operations and state relationships of devices, in order to achieve a certain state of a certain device with the goal of the plant state determined in S22,
Generate what state the devices in the connection relationship should be in (subgoal), and identify the related machine operations that will affect it. In 323, an operation that satisfies a certain requirement could be identified from the static relationship of devices.

However, in order to actually take action as is, it is necessary to propagate it to a dynamic device, perform trial and error, and verify the state of each device. In S24, the predictive simulation function 8 is used to check how the state will change and finally reach the goal state when a series of operations are performed.
By inputting operating parameters to , its state transition is simulated. In S25, the dynamic simulation results and the required state are checked. If this is satisfied, the identified operation sequence is passed to the control action function 4 to deal with the unexpected situation.

If the process is not satisfied and a contradiction occurs, the process returns to S22 and another operation identification is performed again.

Based on the control knowledge newly stored in the control knowledge base 6, the inference unit 5 issues an optimal control command to the plant that flexibly responds to situations of disgrace.

In the above example, a thermal power plant was explained, but
It is clear that the present invention is not limited to this, and can also be applied to general plant control.

The simulations used in the predictive simulation function can be numerical simulations, qualitative simulations (qualitative reasoning), etc. depending on the field of application. Furthermore, in operation identification (S23), in application fields where the sequence is important, a mechanism having sequence generation (operation timing generation) as its function may be considered.

[Effects of the Invention] As explained above, according to the present invention, in contrast to a simple control device that inputs plant status and determines control actions, a knowledge base that stores deep knowledge that is the basis of control knowledge generation is used. The structure is configured to generate new control knowledge based on this deep knowledge and respond when an unexpected situation occurs, so it can respond flexibly to plant conditions and perform optimal control operations safely and stably. It is possible to provide a plant control device that can perform the following operations.

[Brief explanation of drawings]

FIG. 1 is a block diagram of the real side of the plant control system according to the present invention, FIG. 2 is a flowchart showing the processing contents of the inference section 9, and FIG. 3 is a diagram explaining the prior art. 1...Thermal power plant 2...Power plant control device 3...Status monitoring function 4...Control action function 5.9...Inference section 6...Control knowledge base 7
... Countermeasure planning function 8 ... Predictive simulation function 10 ... Deep control knowledge base

Claims (3)

[Claims] (1) Predictive simulation that predicts the future state of a plant in a plant control device equipped with a state monitoring function that detects the state of the plant and a control action function that performs control operations based on experiential knowledge according to the plant state. 1. A plant control device characterized by having a countermeasure planning function that inputs current and future plant conditions and generates control targets and control operations necessary for recovery in the event of an unexpected situation based on deep knowledge. (2) The plant control device according to claim 1, wherein deep knowledge based on a structural model of the plant is employed as the deep knowledge. (3) The plant control device according to claim 1, wherein deep knowledge based on a dynamic characteristic model of the plant is employed as the deep knowledge.