JPH03216705A - Method and device for supporting plant operation - Google Patents

Abstract

PURPOSE:To back up an operation guidance by indexing a past similar example, and overlapping it with a present value trend graph so as to display it when a result value differs from a predicted simulation result. CONSTITUTION:A plant operation support system has a plant 101, a process data input part 102, a process data base 103, a knowledge base 104, an inference processing part 105, an operation guidance 106, a prediction processing part 107, a control model 108, prediction simulation data 109, a feature extraction part 110, a similar example retrieval processing part 111, a result data base 112, similar example data 113, a display editing part 114 and a CRT display device 115. In usual time, operation results up to the present time and the prediction simulation result of the plant by the operation guidance are displayed on a display screen and the similar operation result example retrieved from operation data up to the present is displayed on the same display screen apart from the display at unusual time. Thus, effective operation support information can be supplied to an operator.

Description

Detailed Description of the Invention [Industrial Field of Application] The present invention relates to a plant operation support method and apparatus that supports plant operation using a magazine function. It means to help. A plant operation support device that uses a display screen to support the operation of such a plant is disclosed in Japanese Patent Application Laid-open No. 59-1.
There is No. 16812. JP-A-59-116812 is
Its features include creating operational guidance based on the plant status judgment results, predicting the appropriateness of plant operation according to the created operational guidance, and evaluating the operational guidance. [Question 11 to be solved by the invention] JP-A-59-116812 is an advanced concept in that it evaluates driving guidance. However, there is no detailed description of how to predict or evaluate its validity. Furthermore, even if predictions were to be made, it would be difficult to make predictions that take into account various changes in the plant. On the other hand, predictions generally use simulation models. However, the simulation model is fixed, and even if the model has a certain range, if the plant itself changes significantly, this model will have no meaning, and the simulation results using this model will have large errors. have.

Examples of major changes in the plant itself include plant accidents and abnormalities. In this case, consider it as a countermeasure for accidents or abnormalities. Other than accidents and abnormalities, there are examples of water and sewage systems. A water and sewage system consists of various elements such as an aeration tank, a settling basin, and a drainage pump station, but the models of these elements themselves are relatively simple. However, the model changes greatly depending on the weather. For example, each model will fluctuate greatly depending on whether it is heavy rain, localized torrential rain, or light rain that continues to fall for a long time. However, it is difficult to accurately grasp such a change model. An object of the present invention is to provide a plant operation support method and device that enables appropriate operation in response to changes in the plant without changing the model even when a simulation model is used. ．． [Means for Solving the Problems] The present invention displays on a display screen the operating results up to the present and the prediction simulation results of the plant based on the operation guidance during steady states, and during non-steady states, displays up to the present in addition to the above display. Examples of similar driving results retrieved from driving data will also be displayed on the same display screen. Further, the present invention includes: an input unit that receives process data from a plant; a process database unit that stores the process data from the input unit as performance data; a first means for generating operation guidance using an inference mechanism; a second means for obtaining simulation results when the plant is operated according to the operation guidance; a performance database section that stores past plant performance examples; and during steady state, the operation performance up to the present from the process database section; displaying the simulation results from the second means,
It consists of a display section that also displays, in addition to the above display, examples of similar operation results from the above-mentioned performance database section that have been searched from the operation data up to the present during non-stationary conditions. [Function] According to the plant operation support method and device of the present invention, during steady state, the operating results up to the present and the prediction simulation results of the plant based on the operational guidance are displayed on the display screen, and during non-steady state, other than the above display is displayed. In addition, examples of similar driving results retrieved from the driving data to date are also displayed on the same display screen. [Embodiment] An embodiment of the present invention will be described below with reference to a functional block configuration diagram of a plant operation support system shown in FIG. Measured values of process data serving as operation indicators of the plant 101 are accumulated and stored in the process database 103 via the device and processing of the process data input unit 102.

The driving support system uses this process database 103.
The driving support information is derived based on the C
The information is provided by displaying it on the RT display device 115. IF...THEN driving know-how and general control theory/knowledge gained from past experience is added to the knowledge base 104.
...Stored as driving rules expressed in a format. In the inference processing unit 105. The process values in the process database 103 are checked against the driving rules in the knowledge base 104, and the driving rules that satisfy the IP condition part are executed one after another to derive the driving guidance 106. This driving guidance 106 is displayed on a CRT as one of the driving support information. In addition, as a predictive simulation function, the predictive processing unit 107 uses a control model 10 that simulates a plant.
8 to index and apply the operation guidance 106 and the process values indicating the current operation and process trends from the process database 103 and execute the simulation.
Calculate predicted simulation data 109. This predictive simulation data 109 is also displayed on the CRT as driving support information. Furthermore, it has a function of storing past preferred operating data in the form of actual results in the actual performance database 104 and displaying it on the CRT together with current values and predicted values as similar example data 113. The feature extraction unit 110 indexes time-series process data from the process database 103, develops it into a vector series, and extracts the increase in process value.
Extract the downward change trend as a feature. The similar example search processing unit 111 uses this feature pattern as key data,
Similar example data 113 is searched from the performance database 112. The performance database 112 stores past favorable driving performance data, and similarly manages it using characteristic pattern data expanded into a vector series. Display editing department 1
14 converts the driving guidance l06, predictive simulation data 109, and similar example data 113 together with the process values of the process database 103 into display data, and displays the data on the CRT display device 115 as a message, a trend graph, or the like.

Figure 2 is a hardware configuration diagram of the plant operation support system. The file device 203 stores the process database 103, performance database 112, knowledge base 104, etc. shown in FIG. The CPU 201 executes each processing unit 105, 10 in FIG. 1 according to the program.
7, 110, 111 and file device 203
The driving support information is derived and stored in the main memory 202. main memory 20
2 serves to store processing programs such as search and conversion of process data, data being processed, and driving support information such as driving guidance 106 and similar example data 113. The driving support information stored in the main memory 202 is displayed on the CRT display device 115 by the CRT control device 205. Further, the keyboard 206 and mouse 207 are used for selecting a CRT display screen, switching the display mode of driving support information, and the like. The process input device 204 is
It serves as an external interface with the plant for inputting process data and storing it in file device 203.

FIG. 3 shows an example of a CRT screen for driving support.

Driving guidance display section 301 and trend graph display section 3
It is divided into 02 parts and has a multi-window configuration. The operation guidance display section 301 displays the time at which plant operation is required, as well as its contents, ie, operation instructions and control operations, as messages. On the other hand, the trend graph display section 302 displays past actual values (one hour up to 9:30) and future values based on the current time (10:30) regarding important process data that serves as an index for operation and control operations. The predicted value (1 hour until 11:30) can be displayed as a trend graph. FIG. 3 shows a trend graph 303 for process value 1 and a trend graph 304 for process value 2. In addition, time-series data is displayed on the same time axis for operation charts and trends in control operation amounts.

The trend graph 305 for the control amount 1 shows the past actual value and the future value (set at 50% at 10:40) according to the driving guidance.Similarly, the trend graph 305 for the predicted control amount 2
6 also shows actual values and future predicted values (90% set at 10:50). That is, the solid line trend graph shows the predicted simulation results obtained by operations according to the driving guidance. During normal conditions, driving guidance and predictive simulation graphs can provide driving support, but if actual values differ from predicted values during non-steady conditions, driving decisions may be confused and the information is insufficient as driving support information. . Therefore, when necessary, driving guidance can be backed up by displaying similar data from past favorable driving results at the same time. Similar data trend graphs 307-31
0 is displayed with a broken line in the same color as the actual value/predicted value for each item. By displaying dashed lines in the same color, when the same value is shown at the same time, only the actual value/predicted value is displayed, and only when the value is different, it is displayed, making it easy to recognize it as a driving index. In the example of FIG. 3, in a similar case, the graph 307 for process value 1 will be significantly different from the predicted simulation graph 303 in the future, and the control amount 1 (graph 3
09) was lowered at that timing, which is information that supports driving guidance.

At the bottom left of the trend graph display area 302, place similar display/similarity deletion icons, and use the mouse 207 to display similar performance data 307 to 310 by command/bic operation.
It is possible to display and delete with one touch. FIG. 4 shows the development of time series process data into a vector series. The tendency of change in process amount P at time t is shown as a vector. FIG. 5 shows the vector value conversion of this vector. For example, the magnitudes of vectors 5th Jm(1) to (7) are associated with vector values 3 to -3, and each vector is converted to an approximate vector value. In the example in Figure 4, O
The numerical sequence is -1-0, 1, l, 0, O, -1, -2. The feature extraction unit 110 in FIG. 1 extracts the features of the process data as a numerical string by converting it into a numerical string. The performance database 112 stores past performance process data as well as numerical strings indicating characteristics, and the similar example search processing unit 111 compares the numerical strings and indexes similar examples.
Accumulate the size of the difference between the numerical strings of process data marked with * and past performance data, select performance data with a small value as a candidate for a similar example, and select a similar example by comparing the deviation of the data value of process amount P. do. Plant conditions that require operational support are not steady states but those with large changes in process quantities such as abnormal times, and feature extraction focusing on time-series vectors allows for efficient similar example searches. [Effects of the Invention] According to the present invention, when the actual value differs from the predicted simulation result in driving according to the driving guidance, similar examples are indexed from past suitable driving performance data and the current value trend is calculated. By displaying it over the graph, it can be easily referred to as a driving index, and useful driving support information can be provided to the operator.

[Brief explanation of drawings]

FIG. 1 is a functional block configuration diagram of a plant operation support system according to an embodiment of the present invention, FIG. 2 is a hardware configuration diagram of the plant operation support system, and FIG. 3 is an operation support C
An example of the RT screen, FIG. 4 is a diagram showing how time-series process data is developed into a vector series, and FIG. 5 is a diagram showing how it is converted into a vector value. 101...Plant, 102...Process data input section, 103...Process database, 1
04...Knowledge base, 105...Inference processing unit, 10
6... Driving guidance, 107... Prediction processing unit, l
08... Control model, 109... Prediction simulation data, 110... Feature extraction section, 111... Similar example search processing section, 112... Performance database, 11
3... Similar example data, 114... Display editing department, 11
5...CRT display device, 201...CPU
．． 202... Main memory, 203... File device, 204... Process input device, 205... CR
T control device, 206...Keyboard, 207...Mouse, 301...Driving guidance display section, 302...
・Trend graph display section.

Claims (1)

[Scope of Claims] 1. In a plant operation support method for operating a plant according to the contents displayed on a display screen, during steady state, the display screen displays the operating results up to the present and the predicted simulation results of the plant based on the operation guidance. However, during unsteady conditions, in addition to the above display, examples of similar operation results retrieved from the operation data up to the present are also displayed on the same display screen. 2. The plant operation support method according to claim 1, wherein the operation results in a steady state and the similar operation results in an unsteady state are control amounts to the plant and process values from the plant. 3. An input unit that takes in process data from the plant;
a process database section that stores process data from the input section as performance data; a first means that generates operation guidance by an inference mechanism; and a second means that obtains simulation results when the plant is operated according to the operation guidance. and an actual performance database section that stores past plant performance examples, and during steady state, displays the operating performance up to the present from the process database section and the simulation results from the second means, and during non-steady state, displays the corresponding A plant operation support device comprising: a display section that also displays similar operation performance examples from the performance database section, which are searched using operational data up to the present, in addition to the display; 4. The plant operation support system according to claim 3, wherein the search key is a feature extraction element obtained by converting operational data up to the present into a vector format. 5. The plant operation support system according to claim 4, wherein the operation results in steady state and the similar operation results in unsteady state are control amounts to the plant and process values from the plant. 6. The plant operation support system according to claim 5, wherein when displaying the unsteady state, control amounts to the plant are displayed close to each other, and process values from the plant are displayed close to each other.