JP6885321B2 - Process status diagnosis method and status diagnosis device - Google Patents

Description

The present invention relates to a process state diagnosis method and a state diagnosis device.

There are a model-based approach and a database approach as methods for diagnosing the manufacturing state of the manufacturing process, especially the abnormal state. The model-based approach is an approach in which a model expressing a physical or chemical phenomenon in a manufacturing process is constructed by a mathematical formula, and the manufacturing state of the manufacturing process is diagnosed using the constructed model. On the other hand, the database approach is an approach in which a statistically analytical model is constructed from the operation data obtained in the manufacturing process, and the manufacturing state of the manufacturing process is diagnosed using the constructed model.

In a manufacturing process such as a steel process, many kinds and sizes of products are manufactured on one production line, so that there are many operation patterns. Further, in a manufacturing facility such as a blast furnace, natural objects such as iron ore and coke are used as raw materials, so that the manufacturing process varies widely. Therefore, when diagnosing the manufacturing state of a manufacturing process such as a steel process, there is a limit to the approach using only the model-based approach.

As a database approach, there is a diagnostic method that creates a database of operation data at the time of past abnormalities to determine the similarity with the current operation data, and conversely, creates a database of normal operation data and differs from the current operation data. There is a diagnostic method to determine. However, in a manufacturing process such as a steel process, the number of equipment used for manufacturing is large, and especially when there are many aging equipment such as Japan, unprecedented troubles often occur in the past. .. Therefore, there is a limit to the prediction of an abnormal state by a diagnostic method such as the former based on past trouble cases.

On the other hand, as the latter diagnostic method, there are those described in Patent Documents 1 to 4. Specifically, Patent Documents 1 and 2 describe a method of predicting or detecting an abnormal state of a manufacturing process based on a prediction by a model created using normal operation data. Further, in Patent Documents 3 and 4, patterns are extracted from normal operation data and made into a library, and the difference between the acquired operation data and the library pattern is determined to detect a situation different from usual at an early stage. How to do it is described.

International Publication No. 2013/011745 Japanese Patent No. 4922265 Japanese Patent No. 5651998 Japanese Patent No. 5499900

However, in a manufacturing process such as a steel process, it is difficult to construct a model that completely covers a large number of operation patterns and variations in the manufacturing process. Therefore, the model is used to accurately determine the manufacturing state of the manufacturing process. It is difficult to predict. Therefore, it is difficult to improve the accuracy of predicting or detecting an abnormal state in the manufacturing process by the methods described in Patent Documents 1 and 2. Further, in the methods described in Patent Documents 3 and 4, when there are many operation patterns in the manufacturing process, the number of patterns and the distribution state thereof increase with each operation, so that the acquired operation data and the library are created. A lot of labor is required to determine the difference from the pattern, and it becomes difficult to make a diagnosis efficiently. Therefore, it has been expected to provide a technique capable of accurately and efficiently diagnosing the manufacturing state of the manufacturing process.

The present invention has been made in view of the above problems, and an object of the present invention is to provide a process state diagnosis method and a state diagnosis device capable of accurately and efficiently diagnosing a process state. ..

In the process state diagnosis method according to the present invention, the state of the process is determined by using a deviation index from the normal state of the process calculated for each submodel from the prediction errors of a plurality of submodels that predict the state of the process. It is a state diagnosis method of the process of diagnosis, in which a variable attribute label indicating a variable attribute is defined for each deviation index calculated for each submodel, and an aggregate value is calculated for a deviation index having the same variable attribute label. An attribute deviation index calculation step in which a pattern having the aggregated value of the deviation index for each attribute as an element is stored in the database as an attribute deviation index pattern, and an attribute deviation index pattern calculated from the actual value of the newly obtained variable. It is characterized by including a state diagnosis step of extracting a similar attribute deviation index pattern from the database and presenting past information about the extracted similar attribute deviation index pattern.

In the state diagnosis method of the process according to the present invention, in the above invention, when the number of elements constituting the attribute deviation index pattern is m, each attribute deviation index pattern is set as a point on the m-dimensional space. Similar attributes based on the distance between the points corresponding to the attribute deviation index pattern calculated from the actual values of the newly obtained variables and the points corresponding to the attribute deviation index patterns stored in the database. It is characterized by including a step of extracting a deviation index pattern.

The state diagnosis method of the process according to the present invention is characterized in that, in the above invention, the state diagnosis step includes a step of presenting past information regarding similar attribute deviation index patterns in a ranking format in descending order of similarity. ..

According to the method for diagnosing the state of the process according to the present invention, in the above invention, the aggregated value of the deviation index is a statistic including the average value or the maximum value of the deviation index for the deviation index having the same variable attribute label. It is a feature.

The process state diagnostic apparatus according to the present invention determines the state of the process by using a deviation index from the normal state of the process calculated for each submodel from the prediction errors of a plurality of submodels that predict the state of the process. It is a state diagnostic device for the process of diagnosing, and a variable attribute label indicating the variable attribute is defined for each deviation index calculated for each submodel, and an aggregate value is calculated for the deviation index having the same variable attribute label. An attribute deviation index calculation unit that stores a pattern having the aggregated value of the deviation index for each attribute as an element in the database, and an attribute deviation index pattern calculated from the actual value of the newly obtained variable. It is characterized by including a state diagnosis unit that extracts a similar attribute deviation index pattern from the database and presents past information on the extracted similar attribute deviation index pattern.

According to the process state diagnosis method and the state diagnosis device according to the present invention, the process state can be diagnosed accurately and efficiently.

FIG. 1 is a block diagram showing a configuration of a process state diagnosis device according to an embodiment of the present invention. FIG. 2 is a diagram showing seven deviation index patterns. FIG. 3 is a diagram showing an attribute deviation index pattern in which deviation index patterns are aggregated based on a variable attribute label. FIG. 4 is a flowchart showing a flow of a process state diagnosis process according to an embodiment of the present invention.

Hereinafter, a process state diagnostic apparatus according to an embodiment of the present invention will be described with reference to the drawings. In this embodiment, the present invention is applied to a process of diagnosing a manufacturing state of a manufacturing process of a manufacturing facility such as steel equipment, but the present invention relates to a power generation process of a power generation facility, a transport process of a transport facility, and the like. And the same can be applied to the process of diagnosing the state of other processes such as the waste liquid treatment process of the waste liquid treatment equipment.

〔Device configuration〕
First, a process state diagnostic apparatus according to an embodiment of the present invention will be described with reference to FIG.

FIG. 1 is a block diagram showing a configuration of a process state diagnosis device according to an embodiment of the present invention. As shown in FIG. 1, the process state diagnosis device 1 according to the embodiment of the present invention is a device for diagnosing the manufacturing state of the manufacturing process of manufacturing equipment such as steel equipment, and is an input unit 11, an output unit 12, It includes an external device 13, a storage unit 14, and a control unit 15 as main components. The "manufacturing state of the manufacturing process" to be diagnosed includes the state of the product being manufactured in the manufacturing process.

The input unit 11 is a device that receives actual operation data of a diagnosis target for prediction and cause estimation by a submodel described later via an information / control system network. The input unit 11 inputs the received actual operation data of the diagnosis target to the control unit 15.

The output unit 12 is composed of output devices such as a display device and a printing device, and outputs various processing information of the control unit 15.

The external device 13 is connected to the control unit 15 in a form capable of information communication via a telecommunication line. The external device 13 includes an operation database (operation DB) 13a. In the operation DB 13a, among the time series data of the plurality of types of variables acquired during the past operation of the manufacturing process, the time series data of the plurality of types of variables acquired during the normal operation can be read via the telecommunication line. It is stored in the form. Further, in the operation DB 13a, information on the manufacturing conditions of the manufacturing process at the time of the past operation is also stored in association with the time series data of a plurality of types of variables.

The storage unit 14 is composed of a storage device such as a hard disk device, and is connected to the control unit 15. The storage unit 14 stores a submodel database (submodel DB) 14a and an attribute deviation index pattern database (attribute deviation index pattern DB) 14b. The sub-model DB 14 stores as a sub-model a mathematical formula for calculating a time-series predicted value indicating the state of a product being manufactured in the manufacturing process and / or the manufacturing process. The attribute deviation index pattern DB 14b stores the attribute deviation index pattern having the attribute deviation index of each submodel as an element. Details of the submodel, deviation index, and attribute deviation index pattern will be described later.

The control unit 15 is composed of an arithmetic processing unit such as a CPU (Central Processing Unit), and controls the operation of the entire process state diagnosis device 1. The control unit 15 functions as an attribute deviation index calculation unit 15a and a state diagnosis unit 15b when the arithmetic processing unit executes a computer program. The functions of each of these parts will be described later.

[Submodel configuration]
Next, the configuration of the submodel stored in the submodel DB 14a will be described.

In the present embodiment, the submodel means a mathematical formula for calculating a time-series predicted value indicating the state of a product being manufactured in the manufacturing process and / or the manufacturing process. By constructing a plurality of types of submodels, it is easier to detect an abnormal state at an early stage and estimate the cause than to construct one model in the entire manufacturing process. Here, the known mathematical formula can be used as a submodel for the state of the product being manufactured in the manufacturing process and / or the manufacturing process in which a known mathematical formula for obtaining the predicted value of the time series exists.

On the other hand, for the manufacturing process and / or the state of the product being manufactured in the manufacturing process for which there is no mathematical formula for obtaining the predicted value of the time series, statistical analysis is performed using a plurality of variables other than itself acquired during normal operation of the manufacturing process. The regression equation obtained by the above processing is used as a submodel. In addition, each submodel is given a reliability (a value that increases as the prediction error decreases) according to the prediction error (difference value between the predicted value and the actual value) of the submodel in a predetermined evaluation period. Has been done. If the calculated reliability is low, it is desirable to review the submodel configuration.

[Structure of attribute deviation index pattern]
Next, the configuration of the attribute deviation index pattern will be described with reference to FIGS. 2, 3, and 1.

In the present embodiment, the deviation index means a difference value (prediction error) between the actual value of the variable acquired during normal operation and the predicted value of the variable calculated from the corresponding submodel, and is referred to as the deviation index pattern. Means a pattern that includes the deviation index obtained from each submodel as an element. 2 (a) to 2 (g) show seven deviation index patterns. In FIG. 2, the horizontal axis represents the submodel number (1 to 140), and the vertical axis represents the value of the deviation index. Table 1 below shows an example of a submodel when applied to a finish rolling mill in a steel process.

Since a manufacturing process such as a steel process handles a large number of operation patterns, manufacturing processes, multiple facilities, and a large number of sensor measurements, the deviation index patterns vary widely as shown in FIGS. 2 (a) to 2 (g). Become. Therefore, the calculated deviation index pattern is not necessarily similar to the pattern calculated in the past, and in this case, it is not possible to present information such as the equipment state and the cause of the abnormality.

Therefore, in the present embodiment, a variable attribute label indicating the attributes of variables such as measurement items and target equipment is defined for each variable predicted by the submodel, and the average value or the maximum value of the deviation index is defined for the variables having the same variable attribute label. Etc. are calculated, and the attribute deviation index pattern with the aggregated value of the deviation index for each variable attribute label as an element is calculated. In the case of a finish rolling mill in a steel process, variable attribute labels as shown in Table 1 can be defined. In this case, the attribute 1 is defined as an attribute for identifying each finishing rolling mill, and the attribute 2 is assigned an attribute representing the measurement item predicted by the submodel. As shown in attribute 3 of Table 1, it is also possible to assign attributes that are further classified into measurement items. Here, the aggregated value is not limited to the average value and the maximum value, and statistics such as the median, the mode value, the weighted average, the minimum value, the maximum value of the absolute value, and the minimum value should be appropriately selected and used. Can be done.

Figure 3 shows the effect of aggregation by variable attributes. Seven deviation index patterns (a) to (g) are shown on the left side of FIG. 3, which are the same as those shown in FIG. For each of the deviation index patterns (a) to (g), the attribute deviation index pattern obtained by aggregation by the attribute 2 (attribute representing the measurement item) shown in Table 1 is shown on the right side of FIG. The horizontal axis of the attribute deviation index pattern indicates the identification number (1 to 20) assigned to the variable attribute label defined in attribute 2, and the vertical axis indicates the value of the deviation index aggregated by the maximum value for each variable attribute. .. Since the attribute deviation index patterns aggregated in the measurement items are similar, it can be considered that the events (a) to (c) are caused by the same factors although they are detected by different finish rolling mills. .. The same applies to the events (d) and (e) and the events (f) and (g).

In the present embodiment, the manufacturing state of the manufacturing process is diagnosed by comparing the attribute deviation index pattern stored in the attribute deviation index pattern DB 14b with the attribute deviation index pattern obtained from the actual value of the newly acquired variable. To do. FIGS. 2 (a) to 2 (c) take the same abnormal state for different finishing rolling mills as an example. For example, when the event (a) is stored in the attribute deviation index pattern DB 14b as an abnormal event measured in the past, it is new. It is possible to identify the event (b) and the event (c) acquired in the above as similar events.

When considering the application example of the steel process shown in Table 1 to the finish rolling mill, it is possible to identify the finish rolling mill in which the abnormality has occurred by comparing the attribute deviation index pattern aggregated by attribute 1 with the past cases. By using the attribute deviation index pattern aggregated by attribute 2, it is possible to compare with the past cases by using the measurement items related to the abnormality more specifically as a key. When the past similar events cannot be identified by the attribute deviation index pattern by the attribute 2, it is possible to investigate the past abnormal events of the same system by comparing the attribute deviation index patterns aggregated by the attribute 3. It is the case that the most specific abnormal information can be obtained by identifying similar events in the past in all the attribute deviation index patterns of attributes 1 to 3, but on the other hand, even in some attributes, it is different from the past abnormal events. If similar information is presented, it can be effective in stabilizing the equipment.

[Status diagnosis processing]
Next, with reference to FIG. 4, a process state diagnosis method by the process state diagnosis device 1 according to the embodiment of the present invention will be described.

FIG. 4 is a flowchart showing a flow of a process state diagnosis process according to an embodiment of the present invention. The flowchart shown in FIG. 4 starts at the timing when the input unit 11 inputs the actual operation data to the control unit 15, and the state diagnosis process proceeds to the process of step S1.

In the process of step S1, the attribute deviation index calculation unit 15a reads the actual value data of a plurality of types of variables acquired from the manufacturing process at the processing target time from the operation DB 13a. As a result, the process of step S1 is completed, and the state diagnosis process proceeds to the process of step S2.

In the process of step S2, the attribute deviation index calculation unit 15a uses the data of the actual values of the plurality of types of variables read in the process of step S1 to change the state of the manufacturing process at the processing target time to the manufacturing process during normal operation. A value indicating how different it is from the state of is calculated for each submodel as a deviation index. Specifically, first, the attribute deviation index calculation unit 15a reads the data of the submodel from the submodel DB14a and substitutes the data of the actual value of the variable into the corresponding submodel to predict the processing target time for each variable. Calculate the value. Next, the attribute deviation index calculation unit 15a normalizes the actual value and predicted value data of a plurality of types of variables in order to standardize the difference in the absolute quantity and the unit between the variables. Then, the attribute deviation index calculation unit 15a calculates for each submodel using the difference value between the normalized predicted value of the variable and the normalized actual value at the processing target time as the deviation index. If the reliability of each submodel is different, it is desirable that the deviation index is a value that takes into account the reliability of the submodel. Specifically, the value of the prediction error is updated every time the predicted value of the variable is calculated, and the reliability (for example, 1 / (normalization prediction error of the latest section + 1). The current normalization prediction error (deviation). It is desirable to multiply the deviation index by a weight proportional to (index) = (current prediction error-average value of prediction error at the time of model creation) / prediction error standard deviation at the time of model creation). Here, the reason for adding 1 to the prediction error is to prevent the weight from becoming a very large value when the prediction error is close to zero. Further, the calculation formula of the reliability is an example, and the content of the scenario and the calculation formula may be appropriately changed according to the manufacturing process. As a result, the process of step S2 is completed, and the state diagnosis process proceeds to the process of step S3.

In the process of step S3, the attribute deviation index calculation unit 15a is the same for each deviation index calculated in the procedure of step S2 based on the variable attribute label indicating the attribute of the variable such as the measurement item and the target equipment defined in advance. The aggregated value such as the average value or the maximum value of the deviation index is calculated for the variable having the variable attribute label, and the attribute deviation index pattern having the aggregated value of the deviation index as an element is calculated. As a result, the process of step S3 is completed, and the state diagnosis process proceeds to the process of step S4.

In the process of step S4, the state diagnosis unit 15b has the attribute deviation index pattern DB 14b similar to the attribute deviation index pattern having the aggregated value of the deviation index based on the variable attribute label calculated in the process of step S3 as an element. Determine if it is stored in. Specifically, when the number of elements (= number of variable attribute labels) of the attribute deviation index pattern is m, the state diagnosis unit 15b regards each attribute deviation index pattern as one point in the m-dimensional space, and in step S3. The distance between the point corresponding to the attribute deviation index pattern calculated in the process and the point corresponding to the attribute deviation index pattern stored in the attribute deviation index pattern DB 14b is calculated. Then, when the attribute deviation index pattern whose calculated distance is equal to or less than a predetermined value is stored in the attribute deviation index pattern DB 14b, the state diagnosis unit 15b stores a similar attribute deviation index pattern in the attribute deviation index pattern DB 14b. Determine that it is stored.

As a result of the determination, when a similar attribute deviation index pattern is stored in the attribute deviation index pattern DB 14b (step S4: Yes), the state diagnosis unit 15b advances the state diagnosis process to the process of step S5. On the other hand, when a similar attribute deviation index pattern is not stored in the attribute deviation index pattern DB 14b (step S4: No), the state diagnosis unit 15b ends a series of state diagnosis processes.

In the process of step S5, when the distance calculated in the process of step S4 is equal to or less than a predetermined value, the state diagnosis unit 15b adds and stores the abnormality in the attribute deviation index pattern DB 14b to the attribute deviation index pattern. The abnormal attribute label representing the information such as the cause of the above is read, and the read information is output to the output unit 12. If the abnormal attribute label is defined in a format corresponding to the variable attribute label (in the case of the example shown in Table 1, attributes 1 to 3), the past trouble information obtained can be easily organized. At this time, the state diagnosis unit 15b may output information based on the abnormal attribute label in a ranking format in the order of attribute deviation index patterns having high similarity (= short distance). The operator determines the manufacturing state of the manufacturing process to be evaluated based on the past information output to the output unit 12. As a result, the process of step S5 is completed, and the series of state diagnosis processes is completed.

As is clear from the above description, in the state diagnosis process according to the embodiment of the present invention, the state diagnosis unit 15b has an attribute deviation similar to the attribute deviation index pattern calculated from the actual value of the newly obtained variable. When the index pattern is stored in the attribute deviation index pattern DB 14b, information related to the past is presented for a similar attribute deviation index pattern. That is, after once obtaining the deviation index from the plurality of models created at the normal time, the plurality of deviation indexes are made into a library as the attribute deviation index pattern, and the manufacturing state of the manufacturing process is determined based on the libraryized attribute deviation index pattern. Therefore, the manufacturing state of the manufacturing process can be accurately diagnosed. In addition, even in a manufacturing process that exists in a large number of operation patterns and equipment, it becomes easy to efficiently analyze the attribute deviation index pattern accumulated in the library, so that the manufacturing state of the manufacturing process can be efficiently diagnosed. be able to.

Although the embodiment to which the invention made by the present inventor is applied has been described above, the present invention is not limited by the description and the drawings which form a part of the disclosure of the present invention according to the present embodiment. That is, other embodiments, examples, operational techniques, and the like made by those skilled in the art based on the present embodiment are all included in the scope of the present invention.

1 Process status diagnostic device 11 Input unit 12 Output unit 13 External device 13a Operation database (operation DB)
14 Storage unit 14a Submodel database (submodel DB)
14b Attribute deviation index pattern database (attribute deviation index pattern DB)
15 Control unit 15a Attribute deviation index calculation unit 15b Condition diagnosis unit

Claims (5)

A process state diagnosis method for diagnosing the state of the process using a deviation index from the normal state of the process calculated for each submodel from prediction errors of a plurality of submodels for predicting the state of the process.
For each deviation index calculated for each submodel, a variable attribute label indicating the attribute of the variable predicted by the submodel is defined, and one aggregate value is calculated for a plurality of different deviation indexes having the same variable attribute label. , An attribute deviation index calculation step for storing a pattern having the aggregated value of the deviation index for each attribute as an element in the database as an attribute deviation index pattern.
A state in which an attribute deviation index pattern similar to the newly obtained attribute deviation index pattern calculated from the actual value of the variable is extracted from the database, and past information on the extracted similar attribute deviation index pattern is presented. Diagnostic steps and
A method for diagnosing the state of a process, which comprises. The state diagnosis step is calculated from the newly obtained actual values of the variables by regarding each attribute deviation index pattern as a point in the m-dimensional space, where m is the number of elements constituting the attribute deviation index pattern. It is characterized by including a step of extracting a similar attribute deviation index pattern based on the distance between the point corresponding to the attribute deviation index pattern and the point corresponding to the attribute deviation index pattern stored in the database. The method for diagnosing the state of the process according to claim 1. The state diagnosis method of the process according to claim 1 or 2, wherein the state diagnosis step includes a step of presenting past information about similar attribute deviation index patterns in a ranking format in descending order of similarity. One of claims 1 to 3, wherein the aggregated value of the deviation index is a statistic including the average value or the maximum value of the deviation index for the deviation index having the same variable attribute label. The method for diagnosing the state of the process described in. A process state diagnostic device that diagnoses the process state using a deviation index from the normal state of the process calculated for each submodel from prediction errors of a plurality of submodels that predict the process state.
For each deviation index calculated for each submodel, a variable attribute label indicating the attribute of the variable predicted by the submodel is defined, and one aggregate value is calculated for a plurality of different deviation indexes having the same variable attribute label. , An attribute deviation index calculation unit that stores a pattern having the aggregated value of the deviation index for each attribute as an element in the database as an attribute deviation index pattern.
A state in which an attribute deviation index pattern similar to the newly obtained attribute deviation index pattern calculated from the actual value of the variable is extracted from the database, and past information on the extracted similar attribute deviation index pattern is presented. Diagnosis department and
A process state diagnostic device comprising.

Images