JPH0552606A - Plant monitoring system - Google Patents

Abstract

PURPOSE:To reduce a burden on an operator and also reduce individual differ ence of read-out by providing a process condition detecting means, historical data storing means, historical data extraction means, character coding means, and an output means for outputting expressions by character row. CONSTITUTION:A process condition detecting means 2 detects, as the amount of process conditions indicating the process conditions of a plant to be monitored, for example, a process condition value, target value, and the amount of process control. The amount of process conditions is stored in a historical data storing means 12 in time series. The time-series data for the stored amount of process conditions are read out by a historical data extraction means 13. A character coding means 4 transforms variable patterns to be re-generated by the extraction means 13 into expressions by character row. The amount of process conditions transformed in character row expressions is, for example, displayed, printed, and output, by an output means 6. Thus burden on an operator can be reduced, and also individual difference in read-out of change which is a sign of abnormality can be reduced.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a plant monitoring system, and more particularly to a plant monitoring system which can reduce an operator's burden of monitoring and can detect anomalies with less influence of individual differences.

[0002]

2. Description of the Related Art In plants such as chemical plants,
Since plant process fluctuations have a large impact on the final product, are process trends and changes monitored, anomaly monitoring of the instruments and sensors that make up the control loop, and is the process being operated according to operating patterns? It is necessary to monitor whether or not.

This type of process monitoring is performed by using, as its elements, upper and lower limits, rate of change, deviation and the like regarding data representing the state of the process. The operator of the plant operates the plant depending on whether or not a momentary alarm is generated on a monitoring or control cycle unit for the plant.

However, in many cases, these alarms are already late after they occur. Therefore, it is necessary to catch the precursor of abnormalities in the plant and take measures for avoiding abnormal situations and accidents in advance.

Therefore, conventionally, when operating a plant,
In addition to these alarm processes, in parallel with this, trend graphs of various process data are displayed on a display device to allow operators to monitor them. That is, the operator operates the plant by constantly monitoring these trend graphs to detect the state change and change tendency of the process. This trend graph monitoring is a very important job for the operator especially when the state of the plant changes greatly at the time of plant startup, grade change, shutdown, etc.

[0006]

Nowadays, the operation of a plant is changed from the operation by an operation panel to the operation by operating an input / output device having a display device such as a CRT. Therefore, in order to grasp the state of the plant, the plant operator must perform the CRT operation and monitor dozens of trend screens.

In addition, the number of plant operators has been reduced due to the change to monitoring using a CRT display device. Therefore, in the conventional technique, in order to grasp the state of the plant, the operator must frequently perform the CRT operation with a small number of personnel and monitor dozens of trend screens. This work requires a great deal of labor and tension for the operator for a long time. For this reason, in the conventional plant monitoring system, the burden on the operator becomes heavy in combination with the decrease in the number of operators.

Further, in order to catch the precursor of the plant abnormality, what kind of process change has occurred in the past,
Need to read from trend graph. In this case, whether or not to judge the process change on the trend as an abnormal factor depends on the individual difference of the operator. However, in the conventional technology, the consideration of this point is not taken into consideration, and the reading of the change that is a precursor of the abnormality from the trend graph is judged by the subjective opinion of each operator.

An object of the present invention is to reduce the burden on an operator who monitors a plant and to reduce individual differences in reading changes that are predictive of an abnormality.
To provide a plant monitoring system.

[0010]

According to a first aspect of the present invention to achieve the above object, a process state detecting means for detecting a process state quantity representing a process state of a monitored plant, and the process state detecting means. A historical data storage means for accumulating the process state quantity from the detection means in time series, and a process state quantity from the historical data storage means at a designated sampling cycle from a designated time to the past in response to a start instruction. And a historical data extracting means for regenerating the change pattern,
A plant monitoring system comprising: characterizing means for converting a change pattern by the historical data extracting means into an expression by a character string; and output means for outputting an expression by the character string converted by the characterizing means. Provided.

The above-mentioned historical data extraction device can be activated at a fixed cycle, and a fixed-cycle monitoring device can be provided as a device therefor. This fixed-cycle monitoring device extracts a process state value, a target value, and a process operation amount of a historical data storage device at a fixed time interval in the past from the time when the historical data extraction device is operated at a fixed time interval.

According to the second aspect of the present invention, the process state detecting means for detecting the process state quantity representing the process state of the monitored plant and the process state quantity from the process state detecting means are temporarily stored. A process state storage means for accumulating data, a process data extraction means for extracting a process state quantity from the process state storage means as a change pattern, a characterizing means for converting the extracted change pattern into a character string representation, and the character A plant monitoring system is provided, comprising: output means for reporting the plant status to an operator by means of a character string representation converted by the converting means.

[0013]

In the plant monitoring system according to the present invention,
The process state detecting means detects, for example, a process state value, a target value, and a process operation amount as the process state quantity representing the process state of the monitored plant. These process state quantities are accumulated in time series by the historical data storage means.

The time-series data of the process state quantity accumulated in the historical data storage means is read by the historical data extraction means. That is, this historical data extraction means, for example, according to a start instruction from an operator or a start instruction from a fixed cycle monitoring device to start in a fixed cycle, from a specified time to the past, to a specified sampling cycle. The process state quantity is extracted from the historical data storage means and the change pattern thereof is regenerated. This makes it possible to provide the past process state as inquiry information on a wide time scale (hour, day, day). The extraction condition designation for the historical data extraction means, that is, the time range to be extracted and the sampling period can be designated in advance or each time.

The characterizing means converts the change pattern regenerated by the historical data extracting means into a character string representation. That is, the feature amount of the change pattern of the process state is extracted, and the change state of the process is converted into a character string representation using this feature amount. This makes it possible to identify the state of the process close to that of a skilled operator, regardless of the operator's individuality.

The process state quantity converted into the character string representation is output, for example, displayed and / or printed by the output means. As a result, it is possible to reduce the burden on the operator who monitors the plant and reduce the individual difference in reading the change that is a precursor of the abnormality.

Further, according to the second aspect of the present invention, the process state quantity to be input is temporarily stored and held, and this is sequentially extracted as a change pattern and converted into a character. You can know in real time by expression.

[0018]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 1 shows the configuration of an embodiment of the plant monitoring system of the present invention.

The plant monitoring system shown in FIG.
A process state detector 2 for converting and outputting a process state quantity (process state value, target value and process operation amount) of a monitored process 1 into a process state detection value of a type handled by a monitoring device, and the process state detector Process state storage device 3 for storing process state quantity from 2
And a historical data collection device 11 for collecting the process state quantities stored in the process state storage device 3 in time series, and historical data for storing the process state quantities collected by the historical data collection device 11 for a long period of time. Storage device 12 and the historical data storage device 1
2, the process state quantity stored as the historical data is extracted from the designated arbitrary time at an arbitrary sampling time, and the change pattern is regenerated, and the change of the extracted process state quantity is performed. Representation of pattern by character string (character string expression)
A characterizing device 4 for converting into a character string, a characterized data buffer 5 for temporarily storing a character string expression, and an input / output device 7 having a function of outputting the current and past character string expressions and a function of inputting various instructions. The state conversion device 6 for converting the character string representation stored in the characterized data buffer 5 into a format suitable for the output function of the input / output device 7, and the historical data extraction device 13 are operated in a constant cycle. A fixed-cycle monitoring device 14, a storage device 15 that stores the extraction conditions of the historical data extraction device 13 and parameters for characterizing the characterizing device 4, and the latest from the process state storage device 3 according to an instruction. And a process data extraction device 16 for sequentially extracting data.

A process state storage device 3 for storing the process state quantity from the process state detector 2, a historical data collecting device 11 for collecting the process state quantity in time series, and a historical data for storing the collected historical data. The data storage device 12 constitutes a historical storage means. The historical data storage device 12 is
For example, the process state quantity for each minute is historically edited and saved for the past several years in a time-controlled state. In this case, an area for storing data from the present time to the near past, for example, 5 minutes before the present time, and an area for storing data past this are provided separately. The data from the present to the near past is moved to an area for storing the far past data after a lapse of a fixed time, for example, 5 minutes. By doing so, the data from the present to the near past can be quickly extracted.

The plant monitoring system of this embodiment can be configured by a hardware system as shown in FIG. 2, for example. That is, the plant monitoring system according to the present embodiment includes the main storage device 31 and the central processing unit 32.
A disk device 33, a process input / output device 36,
A dispersion instrumentation system 37, a CRT device 34, a printing device 35, and an input device 38 are provided.

The main storage device 31 has an area for storing the operation program of the central processing unit 32, a storage area which functions as the process state storage device 3 and the parameter storage device 15, and the historical data storage device 1.
2, a storage area for temporarily storing data from the present to the near past, and a storage area forming the characterized data buffer 5 are provided. The central processing unit 32 is
By executing the programs stored in the main storage device 31, various functions of the plant monitoring system of this embodiment are realized. For example, the central processing unit 32 realizes the functions of the process state detecting device 2, the characterizing device 4, the historical data collecting device 11, the historical data extracting device 13, the fixed period monitoring device 14 and the state conversion output device 6.

The disk device 33 functions as a storage area for storing distant past data in the historical data storage device 12. The disk device 33 stores a large amount of past data.

The process input / output device 36 is an interface for taking in measurement signals from a sensor group (not shown) in order to detect the state of the process 1 to be monitored. The measurement data from the process input / output device 36 is processed by the central processing unit 32 and converted into information indicating the process state quantity. Therefore, the sensor group, the process input / output device 36, and the central processing unit 32 constitute a process state detecting device.

As means for obtaining the process state quantity, a distributed instrumentation system may be used instead of the process input / output device. The distributed instrumentation system 37 is a controller that controls the plant, and is connected to the central processing unit 32 via a communication line 39. The data indicating the state of the process collected by the distributed instrumentation system 37, that is, the information indicating the process state quantity is sent to the central processing unit 32 via the communication line 39. Here, a system configured by connection with the distributed instrumentation system has been described in consideration of the existence of a plurality of plants, but in the case of a single plant, a single instrumentation system may be used.

Further, the CRT device 34, the printing device 35 and the input / output device 38 function as an input / output device. CR
At T34, a trend graph or the like is displayed. The printing device 35 prints out various data, trend graphs, and the like.

An input / output device 7 is an input / output device such as a CRT device, a printing device, a voice output device and a buzzer, which is the latter stage of the state conversion output device 6.

The characterizing device 4 approximates the input pattern to a polygonal line, converts the polygonal line into a vector sequence forming the line pattern, and compares this with a vector sequence representing a standard pattern to be compared with the polygonal line. Then, by calculating the degree of similarity and the scale between the two, a process change can be performed by using a pattern identification method that enables pattern identification including patterns with similar shapes but different sizes with a small amount of calculation. Convert the pattern into a string representation. That is, the characteristics of the pattern and the character string expression that the operator can easily understand are given to the standard pattern in advance. For example, an expression that represents the state of a process (“up”, “begins to drop”,
When the input pattern is similar to the standard pattern, the character string expression given to the standard pattern corresponding to this pattern is given to the standard pattern. As a result, the process status is represented by a character string.
By using the state change output device 6, this character string expression is output to the input / output device 7 like other upper and lower limit alarms.

FIG. 4 shows an example of a format output to the input / output device. In the figure, the time when the state change is detected, the information name, the information name comment (service name),
It is composed of a state expression and additional data. In the state expression, it is output by approaching the words of the operator. That is, in the example shown in FIG. 4, “begins to rise”,
It is indicated by a character string indicating a specific concept such as "descent" or "constant".

Regarding the processing for converting to the above character string representation,
This will be described with reference to FIG.

FIG. 7 is a block diagram of a characterizing device according to an embodiment of the present invention. In the figure, reference numeral 110 denotes a displayable input / output device, which inputs processing information such as Tag name, item, sampling period, processing target section, and conversion coefficient, and passes it to the input processing unit 130. A data storage device 120 corresponds to the process state storage device and the historical data storage device of FIG.
Time-series pattern data of temperature, pressure, etc.) is stored. 130 is an input processing unit of the characterizing device,
In FIG. 1, it corresponds to a process data extraction device or a historical data extraction device. This is activated by a signal (data) from the input / output device 110, and based on the Tag name, item, sampling period, and processing section information from the input / output device 110, the corresponding pattern data from the data storage device 120, It is extracted from the processing target section at a given sampling cycle and sent to the memory 140.
Reference numeral 150 denotes a filter storage unit, which is a polygonal line approximation processing unit 1
The feature extraction filter used in 70 is stored. The polygonal line approximation processing unit 170 approximates the input pattern stored in the memory 140 with a polygonal line, and converts the input pattern into a series of line segment vectors forming the polygonal line. Reference numeral 180 denotes a dictionary search processing unit, which compares the vector series of the input pattern sent from the polygonal line approximation processing section 170 with the vector series of the standard pattern stored in the standard pattern dictionary 160 to obtain the corresponding candidate vector series. To search. 190
Is a similarity calculation unit that calculates the similarity and scale factor between the vector sequence of the correspondence candidate obtained by the dictionary search processing unit 180 and the vector sequence of the standard pattern. When the scale factor is in a fixed range and the similarity is equal to or higher than a constant value, the standard pattern number, the vector sequence number of the corresponding input pattern, the similarity factor, and the scale factor are stored in the memory 195. The result stored in the memory 195 is displayed on the input / output device 110.

Next, the internal operation of the characterizing device based on FIG. 7 will be described in detail.

FIG. 12 is a flow chart showing the operation of the polygonal line approximation processing section 170. Broadly speaking, the pattern division processing 210 for calculating the unevenness of the pattern by calculating the sum of products of the input pattern f (t) and the feature extraction filter W2 (x) and obtaining the point τi at which the pattern is divided into polygonal lines
And the feature extraction filter W0 (x) to obtain an average value of the data in the vicinity of the dividing points of the pattern, and to obtain an estimated value fe (τp) of the data at the dividing points.
From both-ends estimation processing 230 for obtaining estimated values at both ends from regression lines at both ends of the data section, and vector-series calculation processing 240 for calculating a vector series when the pattern is polygonal-approximated using the dividing points and the both-ends estimation values It is configured.

First, in the pattern division processing 210, input pattern data f (t) (t = 0, 1, ...,
For T), the sum of products calculation using the feature extraction filter W2 (x) stored in the filter storage unit 150

[0036]

[Equation 1]

Thus, the feature amount g2 representing the degree of unevenness
Calculate (t). Here, a is a constant indicating the spread of the filter. FIG. 8B shows the result of product-sum calculation performed on the example of the input pattern shown in FIG. 8A by the feature extraction filter W2 (x) shown in FIG. 9A. By extracting points having positive and negative extreme values whose absolute values are larger than appropriate constants from the calculation result shown in (Equation 1), the dividing point τp (p = 1, 2, ..., K) of the input pattern is calculated. You can ask. In the example of FIG. 8B, τp (p = 1, 2,
..., 9) are division points.

Next, in the estimation processing 220, the sum of products calculation using the feature extraction filter W 0 (x) stored in the filter storage unit 150 at the division point τ obtained by the pattern division processing 210.

[0039]

[Equation 2]

The feature amount g0 (τp) representing the average value is calculated in accordance with. FIG. 9B shows an example of the feature extraction filter W0 (x). Estimated value fe of pattern at division point τp
(Τp) is obtained by using a feature amount g2 (τp) indicating unevenness and a feature amount g0 (τp) indicating an average value.

[0041]

## EQU3 ## fe (τp) = W0 (0) × g0 (τp) + W2 (0) g2 (τp) (Equation 3)

Next, in both-ends estimation processing 230, estimated values at both ends of the input pattern are obtained. Estimated value of the dividing points obtained in the estimation process 220 (τ1, fe (τ1))
And a regression line passing through (τk, fe (τk)),

[0043]

[Equation 4]

[0044]

[Equation 5]

It is calculated by By substituting t = 0 and t = T into the equation (a) of the equation (4) and the equation (b) of the equation (5), respectively, the estimated values fe (0) at both ends of the input pattern are obtained.
(H1 (0) and fe (T) (= h2 (T)) are obtained. FIG. 8C shows the feature points of FIG. 8B and the estimated values at both ends of the input pattern.

Next, in the vector sequence calculation processing 240, the point sequences (0, fe (0)), (τp, fe (τp)) obtained in the estimation processing 220 and the both-ends estimation processing 230.
From (p = 1, 2, ..., K), (T, fe (T)), a vector sequence Ai = (pi, qi) that approximates the input waveform is

[0047]

[Equation 6]

It is obtained by A vector sequence obtained from the example of the point sequence shown in FIG. 8C is shown in FIG.

FIG. 11 is a flowchart showing the operation of the dictionary search processing unit 180. The standard pattern dictionary 160 is
A number is assigned to each standard pattern and stored in numerical order. In addition, a name representing the feature of the vector is given to each vector forming each vector series. For example, depending on the gradient of the vector, "up", "down",
Names such as "equilibrium" can be given.

First, in step 510, the same name as the name given to the standard pattern is given to each vector of the vector series Ai forming the input pattern. Next, "N = 1" is set in step 515 and the search is started.

In the search process 520, a vector series similar to the standard pattern is selected as a correspondence candidate from the vector series of the input pattern. If a correspondence candidate is found, the flow of processing is the transmission processing 53 according to step 525.
Moving to 0, the standard pattern number N and the corresponding candidate vector sequence are sent to the similarity calculation unit 190. If there is no correspondence candidate, it is determined in step 535 whether the search for all the standard patterns has been completed. When the number of standard patterns is set to Nmax and the search for all the standard patterns is not completed, the dictionary number is updated “N = N + 1” in step 536, and the search for the next standard pattern is continued by the search processing 520. It

FIG. 10 is a flow chart for explaining the details of the processing in the search processing 520. First, in order to retrieve the vector of the standard pattern corresponding to the first vector of the input pattern, in step 601, “is =
It is set to 1 ”. Next, in step 602, “j =
1, i = is ”, and the search is started. When the input pattern vector number i is (m + 1),
The search ends in step 603.

The step 61 compares the name of the vector Ai of the input pattern with the name of the vector Sj of the standard pattern.
0, if no match, go to step 615,
A comparison is made with the next vector of input patterns. If they match, step 620 stores the pair of Ai and Sj in temporary storage 600. In step 625, it is determined whether the search has been completed for all the vectors of the standard pattern. If j ≠ n, the process “i =
i + 1, j = j + 1 "is performed and the search for the next vector is continued. When j = n, the process “is = ik + 1” of step 627 is performed, and the search process with the standard pattern is performed from the is-th input vector. here,
ik is the number of the latest input pattern vector in which the name of the standard pattern vector si matches, and is the number of the latest input pattern vector corresponding to Si in the correspondence candidate sequence stored in the temporary storage. ..

In the similarity calculator 190, the similarity S and the scale factor K between the input pattern and the standard pattern are calculated according to the contents of the temporary storage 600 shown in FIG.

[0055]

[Equation 7]

[0056]

[Equation 8]

It is calculated by Where Si (i =
1, 2, ..., M) are standard pattern vector series, Aij
Is a correspondence candidate of the vector series of the input pattern, and Si
And Aij (i = 1, 2, ..., N1) are correspondence candidates. S in the equation 7 is for calculating the correlation coefficient between vector sequences, and is 1.0 if the patterns are completely similar.
Equation 8 is a value (scale) indicating the scale factor when the vector sequence of the input pattern best matches the vector sequence of the standard pattern.
Is. A certain range in which the degree of similarity and the scale factor with respect to the standard pattern obtained by the degree of similarity calculator 190 are set

[0058]

## EQU00009 ## S.gtoreq.Smin and Kmin.ltoreq.K.ltoreq.Kmax (9) are stored in the memory 195, arranged in descending order of similarity.

Next, the operation of this embodiment will be described.
Here, (1) process state quantity input, (2) historical data input characterizing processing, (3) process data online input processing, (4) characterized data output, (5) characterized data Output control, and
Each of (6) parameter setting will be described in this order.

(1) Input of process state quantity The process state detecting device 2 collects a process state quantity indicating a process state from the process 1 to be monitored at a constant cycle. The process state quantity is analog data or digital data representing measurement information output from a meter (flow meter, thermometer, level meter, etc.) used in a plant. In the present embodiment, the process state quantity particularly targets analog data. The process state quantity data is temporarily stored in the process state storage device 3.

(2) Characterizing process by inputting historical data The process state quantity data stored in the process state storage device 3 is read by the historical data collecting device 11 and is collected in time series for each information source. The historical data storage device 12 stores historical data by linking past data in time series.

Next, the historical data stored in the historical data storage device 12, as shown in FIG.
The historical data data extraction device 13 extracts historical data past the time designated by the input / output device 7 in a unit of a predetermined sampling time, and regenerates the process state change pattern.

This change pattern is input to the characterizing device 4. In the characterizing device 4, first, a process of approximating the change pattern by a polygonal line is performed using the above-mentioned formulas 1 to 5, and this is converted into a vector data string using the formula 6. The polygonal line approximation method calculates the unevenness of the input data by calculating the sum of products of the input data f (t) and the feature extraction function W2 (x) registered in advance, and obtains the point τp at which the unevenness is divided into polygonal lines. , The estimated value fe (τp) (p
= 1, 2, 3, ..., K) is obtained. Next, the polygonal line information is obtained by obtaining regression lines of fe (τp) and fe (τp + 1) whose both ends are the estimated value (p = 1, 2, ..., K).

This vector data string is standardized using the standardized parameters stored in the parameter storage device 15. The standardized vector data string is compared with the standard pattern stored in the parameter storage device 15.
That is, the characterizing device 4 searches the standard pattern group stored in the parameter storage device 15, and if there is a standard pattern similar to the input data, the standard pattern group, or a plurality of similar standard patterns is searched. In some cases,
Using Equations 7, 8 and 9, the degree of similarity is obtained, the most similar standard pattern is determined, and the standard pattern is detected. Then, the name (character string expression), which is attached to the standard pattern in advance and represents the characteristic of the standard pattern, is output as the characterized data.

(3) Characterizing process by online input of process data The process data stored in the process state storage device 3 is determined by the process data extracting device 16 as shown in FIG. Is generated as a process state change pattern for input to the characterizing device 4.

This change pattern is input to the characterizing device 4, and similarly to the case described above, a process of approximating the change pattern by a polygonal line is performed and converted into a vector data string. The polygonal line approximation method calculates the unevenness of the input data by calculating the sum of products of the input data f (t) and the feature selection function W2 (x) registered in advance, and obtains τp at which the unevenness is divided into polygonal lines. Measured value fe (τp) (P =
1, 2, 3, ..., K) Obtain. Next, the regression line information is obtained by obtaining the regression lines of f (τp) and f (τp + 1) whose both ends are the estimated value (p = 1, 2, ..., K). This vector data string is standardized using the standardized parameters stored in the parameter storage device 15.

At this time, parameters for the time scale and the data size are applied in the process of standardizing the polygonal line information. This is because it corresponds to different types of input data (temperature data, flow rate data, pressure data, etc.), reactions, and response speeds. That is, for temperature data, which has a slow response speed (that does not change easily), the time scale is increased and the vector information is sparse with respect to the time scale. In addition, since the response speed of flow rate and pressure data is fast, the time scale should be small,
Make the vector information dense with respect to the time scale. As a result, it is possible to more accurately give a character expression to data that changes slowly (small) or data that changes rapidly (large).

FIG. 13 shows an application example of the character conversion parameter. In the case of FIG. 7A, since the time scale is small, even a slight change (movement) appears as a standardized vector data string. In the case of FIG. 7B, since the time scale is large, even if the data fluctuates considerably, the standardized vector data string changes little.

As described above, according to the present embodiment, it becomes possible to carry out the standardization in consideration of the property of the data.

The standardized vector data string is compared with the standard pattern stored in the parameter storage device 15. That is, the characterizing device 4 corresponds to the parameter storage device 1
The standard pattern group stored in 5 is searched, and if there is a standard pattern similar to the input data, the standard pattern is obtained. If there are a plurality of similar standard patterns, the similarity is obtained. , The most similar standard pattern is determined, and the standard pattern is detected. Then, the name (character string expression), which is attached to the standard pattern in advance and represents the characteristic of the standard pattern, is output as the characterized data.

(4) Output of Characterized Data The output characterized data is stored in the characterized data buffer 5. The stored data is the state conversion output device 6
And is displayed by the input / output device 7. As a result, a message is reported to the operator so that the operator can know the state of the process.

(5) Output Limitation of Characterized Data By the way, when all the characterized data to be output is displayed and output, the operator has a lot of information to see. Therefore, when the state does not change, it is preferable to limit the output of the characterized data. The state conversion output device 6 is used for this purpose. Here, the operation of the state conversion output device 6 will be described with reference to FIG.

In FIG. 3, the characterized data to be displayed this time is taken out from the characterized data buffer 5 (step 3
01), it is checked whether or not the expression of the characterized data is an expression indicating a ramp tendency of the data (step 30).
2). This can be easily realized by, for example, preparing a table in which the characterized data is classified in advance and searching the table. In the present embodiment, two types of expressions for the ramp tendency are assumed: rising and falling. Of course, it is not limited to this.

If the characterized data at this time is not an expression showing a ramp tendency, that is, if an expression having a meaning such as "constant" other than "rise" or "fall" is detected, the characterized data buffer 5 is searched for The characterized data of is read and it is checked whether or not it is the same (step 303). If they are the same, the characterized data is not output, and the lamp counter (the lamp counter is a counter indicating the number of times the lamp state is continued) is set to 0 (step 310).
This means that if the same state continues, it will not be reported to the operator by output as characterized data, and if the state does not change for a certain period of time, it will report to the operator. To do. It should be noted that the previous characterized data is undefined at the start of operation of this device, so that it may be processed exceptionally as constant.

On the other hand, when the characterized data of this time is not the same as the characterized data of the previous time, the output format described in FIG. 4 is edited to output the characterized data of this time (step 304, step). 305). Then, the lamp counter is set to 0 (step 310).

In step 302, if the current characterized data is a ramp tendency expression, it is checked in the same manner as above whether or not the current expression is the same as the previous characterized data expression (step 305). .. Here, the rise and fall of the ramp are distinguished.

If they are not the same, the lamp counter is set to 0 and the lamp start time is stored (steps 311, 31).
2). This time can be set, for example, by acquiring the current time from a clock inside the device and setting it in a register (not shown).

On the other hand, if they are the same, the lamp counter is incremented by 1.
Is added (step 306). Then, it is checked whether or not the lamp counter has reached the set value (step 30).
7). If the ramp counter has reached the specified value, the output format is edited to output the expression indicating the ramp tendency (step 308). The editing of the output format here is performed to an expression such as "ramp (XX) continues for n minutes". Here, for n minutes, step 31
Calculated from the time 2 and the current time. If the lamp counter has not reached the set value, the characterized data is not output.

In this embodiment, as described above, when the ramp tendency continues, after a certain period of time, the fact that the ramp tendency continues and the duration thereof are displayed. The reason for this will be described below.

FIG. 5 shows an example of a process change pattern. Now, at time X, it is assumed that the detector (flow meter or the like) has failed and the process state value 41 is slightly lower than the actual process state value. As a result, the controller of the control system (not shown) acts to increase the process operation amount in order to approach the target value. But,
Even if the process operation amount is increased due to the detector failure,
The deviation is not resolved. As a result, the process operation amount keeps increasing and reaches the process operation amount upper limit value at time Y. When this point is reached, an alarm is usually output and the operator will notice an abnormality. However, since the process was in an abnormal state continuously, the product of the plant would be in an abnormal state.

As in this case, the process variation that takes a long time is hard to notice only by visually observing the trend graph. On the other hand, as in the present embodiment, when the lamp tendency continues for a certain period of time, it is detected and displayed and output as characterized data, so that the operator can output the alarm before the alarm is output. It becomes possible to notice an abnormal situation.

As described above, in the present embodiment, the process state is represented by the character string data, so it is possible to omit the direct monitoring of the trend graph. Therefore,
The burden on the operator is reduced and individual differences in process monitoring are avoided. In addition, when the process status changes or the lamp input continues for a certain period of time, the process status is represented by character string data so that the operator can monitor only a limited character string. The sufficiency and burden are greatly reduced.

(6) Parameter Setting Next, parameter setting in the parameter storage device 15 will be described with reference to FIG.

The parameters are set by inputting a tag name (TAGNO) representing the data name of the target process and the data type (ITEM: process state value, target value, process operation amount). 34 (see FIG. 2), a window 61 is displayed, and the window 61 is displayed.
The selected tag name, for example, T301, and the data type, for example, PV are displayed inside. Further, in the window 61, in addition to the data name and data type of the target process, a list of characterized data named as the state change of the target process data is also displayed.

On the other hand, in selecting the representation of the characterized data corresponding to such setting, the character string data is selected in the selection window 62 by touching the screen of FIG.
To display all of the character string data displayed in this window, or a fixed limit number (for example, up to 8
This is done by selecting character string data up to After making the selection, the confirmation key 61-A in the parameter setting window 61 is designated. Then, by inputting an arbitrary time, the historical data in the range including the time is converted into a character. As a result, the selected character string data is displayed on the screen 63 together with the trend graph of the time range. By examining the trend graph and the character string data, the operator can check whether or not the process state is expressed by a desired expression.

If such a parameter setting is carried out by a trained operator, the subsequent character string representation of the process state is carried out based on this parameter, so that the trend graph can be monitored by the trained operator. Is done in the same way as is done by. Even an operator with low skill level and low alertness can monitor as well as an experienced operator.

In the above embodiment, the historical data extracting device 13 is activated in response to the instruction from the input device 7,
Historical data in the specified time range is extracted. The extraction of the historical data is not limited to this, and may be performed periodically using the fixed cycle monitoring device 14. That is, the constant cycle monitoring device 14 generates a change pattern of a process state change having a constant time width at every predetermined constant time, and the characterizing device 4, the characterizing data buffer 5, and the state conversion output device 6 make the constant value constant. The input / output device 7 can obtain the state change and tendency of the process that occurred in time. As a result, when the operator is replaced, the next operator can easily know the state change of the process accumulated by the previous operator.

[0088]

As described above, according to the present invention,
While reducing the burden on operators who monitor the plant,
Individual differences in reading changes that signal anomalies can be reduced.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of a plant monitoring apparatus according to the present invention.

FIG. 2 is a block diagram showing a system configuration of hardware for realizing the plant monitoring apparatus of the above embodiment.

FIG. 3 is a flowchart showing the operation of the state conversion output device.

FIG. 4 is an explanatory diagram showing an example of a format output to an input / output device.

FIG. 5 is an explanatory diagram showing an example of a process change pattern.

FIG. 6 is an explanatory diagram showing an example of a parameter setting screen for a parameter storage device.

FIG. 7 is a block diagram showing an example of a configuration of a characterizing device used in the above embodiment.

FIG. 8A is a waveform diagram showing an example of an input pattern,
(B) is a waveform diagram showing an example of a product-sum calculation result of the input signal waveform using a feature extraction filter, (c) is an example of the feature obtained in (b) above and an estimated value at both ends of the input pattern The waveform diagram shown in FIG. 5D is a table showing an example of a vector series obtained from the point sequence shown in FIG.

9A and 9B are waveform charts each showing an example of a feature extraction filter.

FIG. 10 is a flowchart showing details of search processing.

FIG. 11 is a flowchart showing an example of a dictionary search processing operation.

FIG. 12 is a flowchart showing an example of a polygonal line approximation processing operation.

FIG. 13 is an explanatory diagram showing an application example of a characterizing conversion parameter.

[Explanation of symbols]

1 ... Process to be monitored, 2 ... Process state detector, 3 ...
Process state storage device, 4 ... Characterizing device, 5 ... Characterizing data buffer, 6 ... State conversion output device, 7 ... Input / output device, 11 ... Historical data collecting device, 12 ... Historical data storage device, 13 ... Historical data extraction Device, 14 ... Periodic monitoring device, 16 ... Process data extracting device.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Satoshi Oishi 5-2-1 Omika-cho, Hitachi City, Ibaraki Prefecture Hitachi Ltd. Omika Plant, Ltd. (72) Takashi Shimizu 5-2 Omika-cho, Hitachi City, Ibaraki Prefecture No. 1 Incorporated company Hitachi Information Control System (72) Inventor Koichi Kawaguchi 5-2-1 Omika-cho, Hitachi City, Ibaraki Prefecture Incorporated company Hitachi Ltd. Omika Plant (72) Inventor Morimasa Ogawa Ushio Kurashiki City, Okayama Prefecture 3rd Street, Mitsubishi Kasei Co., Ltd. Mizushima Plant (72) Inventor Fumihiko Yamanaka 3, Shiodori, Kurashiki, Okayama Prefecture Mitsubishi Kasei Co., Ltd. Mizushima Factory (72) Inventor, Hitoshi Matsuo 3-Chome, Shiodotsu, Kurashiki, Okayama Mitsubishi Inside the Mizushima Plant of Kasei Co., Ltd. (72) Inventor Katsuya Watanabe 3-Chome, Shiodo, Kurashiki City, Okayama Prefecture Inside the Mizushima Plant of Mitsubishi Kasei Co., Ltd.

Claims (6)

[Claims] 1. A process state detecting means for detecting a process state quantity representing a process state of a monitored plant, a historical data storing means for accumulating the process state quantity from the process state detecting means in time series, and a start instruction. In response to the past, the historical data extracting means for extracting the process state quantity from the historical data storing means at the designated sampling cycle from the designated time and regenerating the change pattern thereof, and the historical data extracting means A characterizing means for converting the change pattern by the means into a character string representation; and an output means for reporting the plant status to an operator by the character string representation converted by the characterizing means. Plant monitoring system. 2. The plant monitoring system according to claim 1, further comprising means for accepting an activation instruction from an operator to said historical data extracting means. 3. The plant monitoring system according to claim 1, further comprising a fixed cycle monitoring device for activating the historical data extracting means at a fixed cycle. 4. The historical data storage means includes a process state storage device for storing the process state quantity from the process state detection means, a historical data collection device for collecting the process state quantity in time series, and the collected historical data. The plant monitoring system according to claim 1, 2 or 3, further comprising a historical data storage device that stores data. 5. A process state detecting means for detecting a process state quantity representing a process state of a monitored plant, a process state storing means for temporarily accumulating the process state quantity from the process state detecting means, and this process. By the process data extracting means for extracting the process state quantity from the state storing means as a change pattern, the characterizing means for converting the extracted change pattern into an expression by a character string, and the expression by the character string converted by the characterizing means. A plant monitoring system comprising: an output unit that reports a plant state to an operator. 6. The process state quantity is a process state value,
A target value and a process operation amount,
4. The plant monitoring system according to 4 or 5.