JP2018081335A - Plant operation data analysis system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a plant operation data analysis system by making use of effective operation data of input items other than invalid data such as defects or abnormal values, even if the invalid data are contained when predicting plant abnormality.SOLUTION: A plant operation data analysis system includes extraction means for extracting operation data of a plurality of items obtained from target equipment and items having valid numerical values in the operation data, analysis means for analyzing the operation data of a plurality of extracted items in combination, and display means for displaying the output of the analysis means, together with information of the extracted items or the items not extracted.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a plant operation data analysis system for monitoring a plant or the like.

  Currently, various industries operate various plants to produce various products. If an abnormality occurs in these plants, the production of the products stops and repair costs are incurred. In addition, the production of the product is stopped, which can cause a large loss. It is practically difficult to completely eliminate plant abnormalities and failures. Therefore, it is necessary to grasp a sign of abnormality as early as possible based on the measured value.

  One example of a highly versatile abnormality prediction technique is ART (adaptive resonance theory). This is because the normal category range in which normal data exists is automatically determined in advance using the learning data in the normal state, and if the diagnostic data obtained during operation is within the normal category, it is normal or outside. This is a technology that outputs an alarm considering that it is abnormal. When learning, the category range is calculated from the data of n input items, and at the time of diagnosis, it is determined whether the data is within the category based on the data of n input items. There was a need to have n data input items at the time of diagnosis, and there was a problem that could not be calculated if even one input data could not be used.

  For such a problem, the technique described in Patent Document 1 is known. The technique described in Patent Document 1 is a system for creating data for plant diagnosis. A plant diagnosis system for diagnosing plant abnormality using any one of adaptive resonance theory, vector quantization, and neural network based on the correlation of the created multidimensional diagnosis data is proposed in this document. And when a measured value shows an abnormal value or it is missing, it describes that it excludes from learning data and processes.

  On the other hand, the technique described in Patent Document 2 is also known. The technique described in Patent Document 2 is a facility state monitoring method for detecting an abnormality based on a time-series sensor signal output from a facility or apparatus. A learning process for creating a normal model based on the selected learning data and a feature vector extracted based on the sensor signal are compared with the normal model to perform abnormality identification. And when the acquired multidimensional time series signal is missing, it is described that such data is deleted.

Japanese Patent Laid-Open No. 2015-230576 JP 2011-070635 A

  By using the techniques described in Patent Documents 1 and 2, it is possible to deal with data loss and abnormal values. However, in order to prevent production stoppage due to equipment failure, an actual plant may be equipped with a plurality of main equipment and used in rotation. In such a case, since the data of the stopped facility cannot be obtained, one of the operation data is always lost. When the process of deleting missing data is executed as in the technique described in Patent Document 1 or Patent Document 2, all data of other input items containing valid numerical values are deleted depending on the case. That is, there is a possibility that a large amount of operation data that cannot be used for predicting an abnormality may appear, which is inefficient. As a result, abnormality prediction cannot be performed, and as a result, there is a possibility that the plant breaks down and must be stopped for a long time.

  The above description is limited to the abnormality prediction for the sake of easy understanding, but there may be an analysis means for analyzing the operation data in addition to the abnormality prediction when the plant is operating. Even in such a case, if missing or abnormal values are included in the input data, it is possible that valid data of other input items may be deleted, which is inefficient.

  The present invention has been made in view of such problems, and the object of the present invention is to enable the validity of other input items even if invalid data such as missing or abnormal values are included in predicting plant abnormalities. It is to provide a plant operation data analysis system that utilizes various operation data.

  The plant operation data analysis system of the present invention combines a plurality of items of operation data obtained from the target equipment, an extraction means for extracting items having valid numerical values from the operation data, and a combination of the extracted items of operation data. And analyzing means for analyzing the information, and display means for displaying information of the extracted item or the item not extracted together with the output of the analyzing means.

  According to the present invention, even if a plant having equipment operating in rotation is not all of the operation data and some missing or abnormal value data is included, the operation data can be further analyzed and predicted abnormally. Can continue appropriately and efficiently.

The figure which shows the structural example of the plant operation data analysis system which concerns on 1st embodiment. An example of a display screen that allows the operator to evaluate the analysis result without misunderstanding. The example of the screen which displays an analysis result in a different graph frame according to the combination of the extracted item. The example of a screen which displays an analysis result in the same graph frame by the legend of a different shape and color according to the combination of the extracted item. The example of the screen which displays an analysis result in the same graph frame by the line of a different shape and color according to the combination of the extracted item. The figure which shows the structural example of the plant operation data analysis system which concerns on 2nd embodiment. The figure which shows the structural example of the plant operation data analysis system which concerns on 3rd embodiment. An example of causal relationship data held in a multi-item causal relationship storage device.

Next, modes for carrying out the present invention (referred to as “embodiments”) will be described in detail with reference to the drawings as appropriate.

<Example 1>
FIG. 1 is a diagram illustrating a configuration example of a plant operation data analysis system according to the first embodiment. The target equipment 02 may be equipment, equipment, facilities, a plant, or a large-scale system as long as it can obtain operation data 04 of a plurality of items. A plurality of items of operation data 04 obtained from the target facility 02 are given to the extraction unit 06. The extraction unit 06 has a function of extracting only items having valid numerical values from the operation data 04 of a plurality of items. Here, “valid numerical value” means that numerical values are not missing in the operation data, and there are no character strings such as “N / A”, “DIV / 0”, “-”, “not measured”, And it means the value when the set upper and lower limit values are not exceeded. The operation data of items having such effective numerical values (that is, the extracted operation data 08) is given to the analysis means 10. The analysis means 10 performs information processing on the extracted operation data 08 and gives the output 12 of the analysis means to the display means 16. Further, the information 14 of the item extracted or not extracted is given from the analysis means 10 to the display means 16. In the display means 16, at the same time as the output 12 of the analysis means, the extracted item or the unextracted item so that it can be understood which item data was used to obtain the result or which item data was not used. Information 14 is displayed.

  The analysis means 10 for analyzing the operation data 04 may be anything as long as it uses a plurality of items of operation data 04 as input, performs some information processing, and outputs the output 12 of the analysis means. For example, the Mahalanobis Taguchi method, adaptive resonance theory, vector quantization, neural network, subspace method, deep learning, and the like are listed as typical analysis means 10. Among these, the Mahalanobis Taguchi method, adaptive resonance theory, vector quantization, and subspace method have anomalous prediction functions.

  In general, the operation means 04 of a plurality of items is premised on the analysis means 10 that all valid numerical values of all items are provided. For example, if only one item of a plurality of items is missing, the analysis is performed. Since it cannot be used, the operation data 04 at that time may not be used for analysis.

  Actually, the target facility 02 is configured by a plurality of devices, and the devices may be used in rotation. In such a case, one of the devices will be suspended. If the analysis means 10 tries to analyze the operation data 04 for such a target facility 02 as the target, the data cannot be analyzed at any time, so the analysis cannot be executed. Such a problem can be solved by adopting the configuration shown in FIG. In FIG. 1, extraction means 06 is provided. If even a part of the operation data 04 obtained from the target facility 02 has an effective numerical value, it is possible to reduce wasted data by analyzing only that value.

  However, if the input items of the extracted operation data 08 given to the analysis unit 10 are different, it is common that the output 12 of the analysis unit cannot be compared side by side. Therefore, a screen as shown in FIG. 2 is displayed on the display means 16 so that the operator can properly evaluate the analysis result without misunderstanding. In the first line of FIG. 2, an input item “input item 2” that is not used is displayed. As indicated by “missing” on the right side, why the input item was not used may be displayed at the same time. Further, instead of the first line in FIG. 2, an input item “used in analysis” may be indicated. The second line of FIG. 2 is information on the output 12 of the analyzing means, and here, as an example, the result of the abnormality prediction calculation is shown. The figure below it is an example of a trend graph, and the dotted line on the right indicates the latest time. For example, when the analysis means 10 has a function of predicting an abnormality, the alarm level is gradually increased or decreased with respect to time by showing information to the operator in the trend graph as shown in this figure. It is possible to convey information that is easy to understand intuitively, such as being there or the situation has not changed much.

  It is conceivable that the input items of the operation data 04 change in a pattern with time. That is, when there are three pumps from Unit 1 to Unit 3 in a plant and each unit is rotating, analysis results when Unit 1 is moving, analysis when Unit 2 is moving As a result, three patterns of analysis results when Unit 3 is moving are assumed. When the respective patterns are combined and displayed in a trend graph as shown in FIG. 2, numerical values that cannot be compared with each other are arranged side by side, which may cause misunderstanding.

  In order to avoid this, for example, as shown in FIG. 3, there is a method of displaying the analysis result in a different graph frame according to the combination of the extracted items. FIG. 3 shows an example in which three types of trend graphs are displayed on one screen. The character string (text) in the upper left indicates the latest situation, and in this example, the data for pumps 2 and 3 are missing at this time. It also shows that the calculation result of abnormality prediction is alarm level 2. The trend graph on the right side shows the transition of the alarm level in the pattern “Data for pumps 2 and 3 is missing”. The dotted line on the right shows the latest time. The place where there is a blank in the plot is the period of the pattern “The data of pumps No. 2 and No. 3 were not missing”. Similarly, the lower left corresponds to the pattern “Data for pumps 2 and 1 were missing”, and the lower right corresponds to the pattern “the data for pumps 1 and 3 was missing”. In this way, when there are three types of patterns, the trend graph is displayed in separate graph frames, so that past results and transitions corresponding to the combination of devices currently operating can be grasped without error. .

  Alternatively, as shown in FIG. 4, there is a method of displaying in the same graph with legends of different shapes and colors depending on the combination of extracted items. FIG. 4 shows an example in which one trend graph is displayed on one screen. Legends with different shapes and colors are assigned to combinations of operating devices. By using such a display, it is possible to grasp past results corresponding to the combination of devices currently operating without error. Furthermore, the number of trend graphs to be displayed can be reduced compared to the case of FIG. 3, and the display screen to be displayed can be used effectively, and at the same time, the selection of a graph to be viewed can be eliminated.

  Alternatively, as shown in FIG. 5, there is a method of displaying in the same graph as lines of different shapes and colors depending on the combination of extracted items. FIG. 5 shows an example in which one trend graph is displayed on one screen. Differently shaped lines are assigned to combinations of operating devices. By using such a display, it is possible to grasp past results corresponding to the combination of devices currently operating without error. Furthermore, the number of trend graphs to be displayed can be reduced compared to the case of FIG. 3, and the display screen to be displayed can be used effectively, and at the same time, the selection of a graph to be viewed can be eliminated. Unlike the case where only the legend is shown as shown in FIG. 4, the trend of increase / decrease becomes easier to grasp by the line corresponding to the combination of operating devices. However, if there are too many combinations of operating devices, the number of displayed lines increases, which may be difficult to see.

  As output, the combination of FIGS. 4 and 5, that is, the legend and the line may be displayed simultaneously. Thereby, there is a possibility that information can be presented to the operator more easily.

  As described above, by taking the first embodiment, it is possible to reduce the waste of the operation data 04 that has been excluded so far because there is no data loss or some valid data. Further, even when the analysis result cannot be evaluated side by side with respect to the data set used in the analysis, it is possible to output information that can be judged without an operator's misunderstanding.

<Example 2>
FIG. 6 is a diagram showing a configuration example of a plant operation data analysis system according to the second embodiment. The target equipment 02 may be equipment, equipment, facilities, a plant, or a large-scale system as long as it can obtain operation data 04 of a plurality of items. A plurality of items of operation data 04 obtained from the target facility 02 are given to the extraction unit 06. A plurality of items of past operation data 22 obtained from the past operation data storage device are also provided to the extraction means 06.

  The extraction unit 06 has a function of extracting only items having valid numerical values from the operation data 04 of a plurality of items. Here, the valid numerical value is missing in the operation data or does not contain character strings such as “N / A”, “DIV / 0”, “-”, “unmeasured”, and setting. It means a value when the upper and lower limit values are not exceeded. Unlike the first embodiment, in the second embodiment, the item having the above-described valid numerical values in the past operation data 22 obtained by the extraction unit 06 from the past operation data storage device 18 is present. The operation data is also extracted.

  The extracted operation data 08 (data extracted from current operation data and past operation data) is given to the analysis means 10. The analysis means 10 performs information processing on the extracted operation data 08 and gives the output 12 of the analysis means to the display means 16. Further, the information 14 of the item extracted or not extracted is given from the analysis means 10 to the display means 16. In the display means 16, at the same time as the output 12 of the analysis means, the extracted item or the unextracted item so that it can be understood which item data was used to obtain the result or which item data was not used. Information 14 is displayed.

  By taking this form, it is possible to use the analyzing means 10 that uses the past operation data 22 as compared with the first embodiment. For example, it is possible to apply the analyzing means 10 that machine-learns the range of normal operation using the past operation data 22 and determines whether it is abnormal or normal by applying the current operation data 20 to it. It becomes.

<Example 3>
FIG. 7 is a diagram illustrating a configuration example of a plant operation data analysis system according to the third embodiment. The target equipment 02 may be equipment, equipment, facilities, a plant, or a large-scale system as long as it can obtain operation data 04 of a plurality of items. A plurality of items of operation data 04 obtained from the target facility 02 are given to the extraction unit 06. A plurality of items of past operation data 22 obtained from the past operation data storage device are also provided to the extraction means 06.

  The extraction unit 06 has a function of extracting only items having valid numerical values from the operation data 04 of a plurality of items. Valid values here are those that are missing in the operation data, or that do not contain character strings such as `` N / A '', `` DIV / 0 '', ``-'', `` not measured '', etc. It means a value when the upper and lower limit values are not exceeded. In addition, the extraction means 06 extracts the operation data of the items having the above-described effective numerical values from the past operation data 22 obtained from the past operation data storage device 18.

  Unlike the first and second embodiments, the causal relationship data 26 is given to the extraction means 06 from the causal relationship storage device 24. An example of the causal relationship data 26 is shown in FIG. Although FIG. 8A is shown as a table for easy understanding, it may actually be given as data in csv format or text format. In this table, “Cause” of “Causality” is shown on the left and “Cause” of “Causality” is shown on the left. This is a table showing whether the left item affects the upper item. It has become. In the table, "0", "1", and "2" are shown as numerical values. Of these, “0” indicates no causal relationship, “1” indicates a causal relationship, and “2” indicates the same item. The operation of the extraction means 06 for the causal relationship data 26 is as follows.

  (1) If the numerical value is “2”, it is ignored because it is the same item.

  (2) If the numerical value is “0”, it means that there is no causal relationship and is ignored.

  (3) If the numerical value is “1”, it means that there is a causal relationship. If the “causal” data is not a valid numerical value, even if the “fruit” data is a valid numerical value, Data is considered indefinite and is not considered a valid number.

  For example, when the causal relationship data 26 in FIG. 8A is set, the obtained data is (flow rate 1, flow rate 2, flow rate 3, total flow rate) = (100, N / A, N / A, 450). Further, it is assumed that the upper limit of the flow rate is set to 1000 and the lower limit is set to 0. First, the extraction means 06 extracts valid data from here. In accordance with the rules described above, the values of “flow 1” and “total flow” that do not contain a character string and are within the upper and lower limits are valid, and “flow 2” and “flow 3” that contain a character string. Judge that the value is not valid. Then, the causal relationship data 26 is referred to. First, reference is made to the first line, but “flow rate 1” is effective, so no processing is performed. Next, refer to the second line. Since “flow rate 2” is not effective, the causal relationship is examined using this table. Since the numerical value “1” is “total flow rate”, the value of “total flow rate”, which is a numerical value of 450 here, is considered to be invalid. Next, referring to the third line, similarly, a process for regarding that the value of “total flow rate” is not valid is performed. Finally, refer to the fourth line. Since “total flow” is processed as not valid, this line is searched. However, since no item with the numerical value “1” is found, the process is terminated. Eventually, the extraction unit 06 selects only “flow rate 1” as an item having a valid numerical value from among the operation data 04 of a plurality of items, and provides the extracted operation data 08 to the analysis unit 10.

  Suppose that the total flow rate = flow rate 1 + flow rate 2 + flow rate 3. Among these, when the flow rate 2 and the flow rate 3 are unknown (N / A), there are an infinite number of combinations. Even if data processing is performed with the total flow rate = 450, the actual operation mode may be completely different between (flow rate 2, flow rate 3) = (0, 450), (225, 225), and (450, 0). If they are collectively recognized as being in the same state and analyzed, problems may occur in the analysis results. When the abnormality prediction function using learning data and diagnostic data is used, the learning data (flow rate 2, flow rate 3) is actually used even if diagnosis is normal because the learning data and diagnosis data are “total flow = 450”. ) = (0, 450) and the diagnostic data is (flow rate 2, flow rate 3) = (450, 0), there is a possibility of failure. Nevertheless, the reality is that it failed.

  On the other hand, when the causal relationship data 26 of FIG. 8A is set, the obtained data is (flow rate 1, flow rate 2, flow rate 3, total flow rate) = (100, 120, 80, N / A) Suppose that Further, it is assumed that the upper limit of the flow rate is set to 1000 and the lower limit is set to 0. First, the extraction means 06 extracts valid data from here. In accordance with the rules described above, it is determined that the values of “flow rate 1”, “flow rate 2”, and “flow rate 3” are valid, and the value of “total flow rate” including a character string is not valid. Then, the causal relationship data 26 is referred to. First, reference is made to the first line, but “flow rate 1” is effective, so no processing is performed. Next, refer to the second line. Since “flow rate 2” is also effective, no processing is performed. Similarly, no processing is performed for “flow rate 3”. Finally, refer to the fourth line, and search for this line because "total flow" is not valid. However, since no item with the numerical value “1” is found, the process is terminated. Eventually, the extraction unit 06 selects “flow rate 1”, “flow rate 2”, and “flow rate 3” as items for which valid numerical values exist among the plurality of items of operation data 04, and extracts the operation data 08 to the analysis unit 10 as the extracted operation data 08. give. In this case, as described above, innumerable combination problems do not occur, and it can be said that evaluation that “actually failed despite being able to analyze data and judged to be normal” does not occur. .

  It is also assumed that the causal relationship data 26 is as shown in FIG. The difference from FIG. 8A is that “1” is included in the lower left cell of the numerical value “2”. This means that the search needs to be performed retrospectively or repeated. As an example, assume that the obtained data is (flow rate 1, flow rate 2, flow rate 3, total flow rate) = (100, N / A, 600, 950). In accordance with the rules described above, it is determined that the values of “flow rate 1”, “flow rate 3”, and “total flow rate” are valid, and that the value of “flow rate 2” including a character string is not valid. Then, the causal relationship data 26 shown in FIG. First, reference is made to the first line, but “flow rate 1” is effective, so no processing is performed. Next, refer to the second line. Since “flow rate 2” is not effective, the causal relationship is searched. Since the numerical value “1” is “flow rate 3”, it is assumed that the value of “flow rate 3” is valid in terms of the upper and lower limits of 600, but this value is regarded as invalid. Apply processing. Next, refer to the third line. Since “flow rate 3” is processed as not valid, this line is searched. Then, it is understood that the numerical value “1” is “flow rate 1”. Therefore, processing is performed that regards “flow rate 1” as not effective. Next, reference is made to the fourth line, but since “1” is not included, no processing is performed. Then go back to the first line again. Since “flow rate 1” has been processed as being ineffective, “1” is entered in “total flow rate” when searching, and therefore, processing is performed assuming that the value of “total flow rate” is not effective. Next, refer to the second line and confirm that no new processing occurs. Similarly, confirm that no new processing occurs in the 3rd and 4th lines. Furthermore, it is confirmed that no new processing occurs again from the first line to the fourth line, and the process ends. As a result, no valid data remains. As described above, when “1” is included in the lower left cell of the numerical value “2”, the search is performed retrospectively or repeatedly.

  Thus, by taking the third embodiment, it is possible to perform analysis with the operation data 08 extracted by excluding items affected by indefinite conditions, and increase the possibility of obtaining more accurate results. be able to.

02 Target equipment
04 Operation data
06 Extraction means
08 extracted operation data
10 Analysis means
12 Output of analysis means
14 Information on items extracted or not extracted
16 Display means
18 Storage device for past operation data
20 Current operation data
22 Past operation data
24 Causal storage device
26 Causal data

Claims (7)

Multiple items of operation data obtained from the target equipment,
An extraction means for extracting items having valid numerical values from the operation data;
An analysis means for analyzing the combined operation data of the extracted items,
Display means for displaying information of the extracted item or the item not extracted together with the output of the analysis means;
A plant operation data analysis system comprising: Current operation data of multiple items obtained from the target equipment,
A storage device that accumulates past operation data of multiple items obtained from the target facility;
An extraction means for extracting items in which valid numerical values exist among the operation data from the current operation data and past operation data;
An analysis means for analyzing a combination of the current operation data and past operation data of the plurality of extracted items;
Display means for displaying information of the extracted item or the item not extracted together with the output of the analysis means;
A plant operation data analysis system comprising: 3. The extraction unit according to claim 1, wherein the extraction unit searches for and extracts items affected by items for which no valid numerical value exists based on causal relationship data obtained from a causal relationship storage device that stores a plurality of items of causal relationships. A plant operation data analysis system characterized by having a function to be excluded from data. 4. The plant operation data analysis system according to claim 1, wherein the analysis unit has an abnormality prediction function for predicting an abnormality of the target facility. 5. 5. The plant operation data according to claim 1, wherein at least one of Mahalanobis Taguchi method, adaptive resonance theory, vector quantization, neural network, subspace method, and deep learning is used as the analysis means. Analysis system. 6. The plant according to claim 1, wherein when the output of the analysis means is displayed as a trend graph, the display means is displayed in a different graph frame according to a combination of extracted items. Operation data analysis system. 7. The display according to claim 1, wherein when the output of the analysis means is displayed as a trend graph as the display means, a legend of a different shape or color, or a different shape or color depending on the combination of extracted items. A plant operation data analysis system characterized by being displayed in a graph as a line.

Images