JP6711323B2 - Abnormal state diagnosis method and abnormal state diagnosis device - Google Patents

Description

The present invention relates to an abnormal state diagnosis method and an abnormal state diagnosis apparatus for a process such as a manufacturing process.

There are a model-based approach and a database approach as methods for diagnosing abnormal states of processes such as manufacturing processes, power generation processes, and transportation processes. The model-based approach is an approach of constructing a model that represents a physical or chemical phenomenon in a process by a mathematical formula and diagnosing an abnormal state of the process using the constructed model. On the other hand, the database approach is an approach in which a statistical analysis model is constructed from the operation data obtained in the process and the abnormal state of the process is diagnosed using the constructed model.

In a manufacturing process such as a steel process, since a single manufacturing line manufactures products of various types and sizes, there are innumerable operating patterns. Further, in a manufacturing process such as a blast furnace, since natural substances such as iron ore and coke are used as raw materials, there are large variations in the manufacturing process. Therefore, when diagnosing an abnormal state of a manufacturing process such as a steel process, there is a limit to the approach based only on the model-based approach.

As a database approach, there is a diagnostic method to make a database of past operation data when abnormalities occur to judge the similarity with the current operation data, and conversely, to make a database of normal operation data and the difference from the current operation data. There is a diagnostic method for determining. However, in manufacturing processes such as steel processes, in addition to the large number of facilities used for manufacturing, especially when there are many aging facilities such as Japan, unprecedented troubles often occur. .. Therefore, the former diagnostic method based on past trouble cases has a limit in predicting an abnormal state.

On the other hand, as the latter diagnostic method, there are methods described in Patent Documents 1 and 2. Patent Documents 1 and 2 specifically describe a method of predicting or detecting an abnormal state of a manufacturing process based on prediction by a model created using operation data at normal times.

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

However, the methods described in Patent Documents 1 and 2 are limited to predicting or detecting the abnormal state of the manufacturing process, and cannot estimate the cause of the abnormal state occurring in the manufacturing process. Therefore, it has been expected to provide a technique capable of estimating the cause of the abnormal state that has occurred in the manufacturing process.

The present invention has been made in view of the above problems, and an object thereof is to provide an abnormal state diagnosing method and an abnormal state diagnosing device for a process capable of estimating the cause of an abnormal state occurring in a process such as a manufacturing process. Especially.

In order to solve the above problems and achieve the object, a method for diagnosing an abnormal state of a process according to the present invention uses a plurality of sub-models for predicting the state of the process to obtain a deviation index from the normal state of the process. Calculated, based on the deviation index pattern set consisting of deviation indexes calculated for each of the sub-models, a method for diagnosing an abnormal state of a process for estimating the cause of an abnormal state occurring in the process, wherein the deviation index pattern set , A base pattern creating step of creating a plurality of base patterns of anomalous patterns shown when there is an anomaly in each of the variables related as explanatory variables of the sub-model; An abnormal cause estimating step of estimating the cause of the abnormal state.

Further, the method for diagnosing abnormal state of the process according to the present invention is, in the above-mentioned invention, creating a plurality of artificial departure index pattern sets by combining a plurality of types of the base patterns created in the base pattern creating step, and creating the artificial departure index patterns. A base pattern learning step of learning the type of the base pattern included in the set, and the base pattern included in the deviation index pattern set obtained during actual operation based on the learning result in the base pattern learning step. And a basis pattern estimating step of estimating the type of the abnormal pattern, wherein the abnormal cause estimating step estimates the cause of the abnormal state occurring in the process based on the type of the basic pattern.

Further, in the above-mentioned invention, the method for diagnosing abnormal state of the process according to the present invention is characterized in that the base pattern creating step is performed on For each variable related as an explanatory variable of the sub model, a perturbation data set is created by adding one minute anomaly to each variable, and the perturbation data set is given as an input to the sub model to determine the basis pattern. It is characterized by creating.

Further, the abnormal state diagnosis method of the process according to the present invention is the above-mentioned invention, wherein the base pattern learning step selects a plurality of the base patterns created in the base pattern creating step, and is suitable for each of the selected base patterns. A plurality of artificial deviation index pattern sets are created by adding and adding different weights, machine learning is performed using the artificial deviation index pattern sets as input, and the type or weight of the selected base pattern as output. Is characterized by.

In order to solve the above-mentioned problems and achieve the object, the abnormal state diagnosis device for a process according to the present invention uses a plurality of submodels for predicting the state of the process to obtain a deviation index from the normal state of the process. Calculated, based on a deviation index pattern set consisting of deviation indexes calculated for each of the sub-models, an abnormal state diagnosis device of the process of estimating the cause of an abnormal state occurred in the process, in the deviation index pattern set , With respect to the variable related as an explanatory variable of the sub-model, a base pattern creating means for creating a plurality of base patterns of anomalous patterns shown when there is an anomaly, respectively, and a base pattern creating means that is generated in the process based on the base pattern. An abnormal cause estimating means for estimating the cause of the abnormal state.

Further, the abnormal state diagnosis device of the process according to the present invention, in the above invention, creates a plurality of artificial deviation index pattern sets by combining a plurality of types of the basic patterns created by the basic pattern creation means, and the artificial deviation index patterns. A basis pattern learning means for learning the type of the basis pattern included in the set, and the basis pattern included in the deviation index pattern set obtained during actual operation based on the learning result in the basis pattern learning means. And a base pattern estimating means for estimating the type of the base pattern. The abnormality cause estimating means estimates the cause of the abnormal state occurring in the process based on the type of the base pattern.

Further, the abnormal state diagnosis device for a process according to the present invention is, in the above-mentioned invention, the base pattern creating means is characterized in that, with respect to a reference data set composed of data of a normal state of the process necessary for calculation of the sub-model, For each variable related as an explanatory variable of the sub model, a perturbation data set is created by adding one minute anomaly to each variable, and the perturbation data set is given as an input to the sub model to determine the basis pattern. It is characterized by creating.

According to the present invention, even if the same trouble has not occurred in the past, it is possible to quickly identify the cause of the abnormal state that has occurred in a process such as the manufacturing process and take measures.

FIG. 1 is a block diagram showing a configuration of a process abnormal state diagnosis device according to an embodiment of the present invention. FIG. 2 is a flowchart showing the overall flow of the abnormal state diagnosis method by the abnormal state diagnosis device for a process according to the embodiment of the present invention. FIG. 3 is a diagram showing an example of a deviation index pattern set created by the abnormal state diagnosis device for a process according to the embodiment of the present invention. FIG. 4 is a diagram showing an example of a base pattern created by the abnormal state diagnosis device for a process according to the embodiment of the present invention. FIG. 5 is a diagram showing an example of the result of inversely estimating the base pattern from the deviation index pattern set in the abnormal state diagnosis device for the process according to the embodiment of the present invention.

Hereinafter, a process abnormal state diagnosing method and an abnormal state diagnosing device according to embodiments of the present invention will be described with reference to the drawings. The present invention is not limited to the embodiments below. In addition, constituent elements in the following embodiments include elements that can be easily replaced by those skilled in the art, or substantially the same elements. In the following description, “A and/or B” means “at least one of A and B”.

[Abnormal condition diagnostic device]
The abnormal state diagnosis device 1 is a device for diagnosing abnormal states of various processes such as a manufacturing process of a manufacturing facility such as a steel facility, a power generation process of a power generation facility, a transportation process of a transportation facility, and the like, as shown in FIG. The input unit 10, the output unit 20, the external device 30, the storage unit 40, the model definition unit 50, the learning unit 60, and the control unit 70 are provided as main components. In the following description, an example in which the abnormal state diagnosis device 1 is applied to a manufacturing process such as a steel process will be described.

The input unit 10 is a device that receives, through an information/control system network, actual operation data of a diagnosis target that performs prediction and cause estimation by a sub model described later. The input unit 10 inputs to the control unit 70 the actual operation data of the diagnosis target received from, for example, a process computer (not shown).

The output unit 20 is configured by an output device such as a display device and a printing device, and outputs various processing information of the control unit 70.

The external device 30 is connected to the model defining unit 50 and the control unit 70 in a form capable of information communication via an electric communication line. The external device 30 includes an operation database (hereinafter referred to as “operation DB”) 31. In the operation DB 31, the actual values of a plurality of types of variables acquired during the past operation of the manufacturing process, that is, the time series data of a plurality of types of variables (hereinafter referred to as “variable data”) are obtained during normal operation. A plurality of types of variable data are stored in a form that can be read via an electric communication line.

The storage unit 40 is configured by a storage device such as a hard disk device, and is connected to the model definition unit 50, the learning unit 60, and the control unit 70. The storage unit 40 includes a sub-model database (hereinafter referred to as “sub-model DB”) 41, a base pattern database (hereinafter referred to as “base pattern DB”) 42, and a learning result database (hereinafter referred to as “learning result DB”) 43. Is remembered.

The sub-model DB 41 stores, as a sub-model, a mathematical formula for calculating a time-series predicted value indicating a manufacturing process and/or a state of a product being manufactured in the manufacturing process. Further, the base pattern DB 42 stores the base pattern for each sub model. The learning result DB 43 stores the learning result of the base pattern.

The model defining unit 50 functions as the sub-model creating unit 51 and the base pattern creating unit 52 when the arithmetic processing device executes the computer program. The functions of these parts will be described later.

The learning unit 60 functions as the base pattern learning unit 61 when the arithmetic processing device executes the computer program. The function of the base pattern learning unit 61 will be described later.

The control unit 70 is configured by an arithmetic processing device such as a CPU (Central Processing Unit) and controls the operation of the entire abnormal state diagnosis device 1. The control unit 70 functions as the deviation index calculation unit 71, the abnormality detection unit 72, the base pattern estimation unit 73, and the abnormality cause estimation unit 74 by the computer processing device executing the computer program. The functions of these parts will be described later.

(Sub model)
In the present embodiment, the sub-model indicates a relationship between, for example, a state of material before manufacturing, a setting of equipment before manufacturing, a state of equipment during manufacturing, a state of a product during manufacturing and/or after manufacturing, and the like. It means a mathematical formula. As the sub-model, for example, a mathematical expression as a forward model for predicting the state of the product during and/or after production from the state of the material before production, the setting state of equipment before production, the state of equipment during production, etc. In addition, as an inverse model that inversely estimates whether the setting of the equipment before manufacturing was appropriate based on the state of the material before manufacturing, the state of the equipment during manufacturing, the state of the product during manufacturing and/or after manufacturing, etc. There are various mathematical models that mutually estimate each other, such as Further, it may be a model in which the state quantity of the power generation equipment or the transportation equipment included in the manufacturing process is estimated from various sensors, other state quantities, and set values. As described above, by constructing a plurality of types of sub-models, it becomes easier to detect an abnormal state earlier and estimate the cause than to construct one model in the entire manufacturing process.

In the manufacturing process, various models have been built in order to create products with the quality and dimensions that are targeted, and it is made to predict the state of the manufacturing process and the state of the product being manufactured. An existing model may be used as a sub model. If the number of sub-models is insufficient, a new sub-model can be added by statistical processing. For example, a regression equation can be obtained using a plurality of variables other than the one acquired during normal operation of the manufacturing process and used as a sub model. In addition, reliability is given to each sub-model according to the prediction error of the sub-model in a predetermined evaluation period (a value that increases as the prediction error decreases). If the calculated reliability is low, it is desirable to review the configuration of the submodel, but it is not possible to review it freely at any time, so for example, within a certain period, even if the reliability is low, it must be used without reviewing. Sometimes it doesn't.

(Deviation index)
In the present embodiment, the deviation index means, for example, a difference value or a ratio between the predicted value calculated from the sub-model and the actual value of the manufacturing process corresponding thereto, or a value calculated based on these. ing. It is more preferable that the deviation index is a value calculated by combining the above-mentioned reliability levels. The combination method in this case may be, for example, “deviation index with reliability consideration”=“deviation index without reliability consideration×reliability”. The deviation index is a value at a timing to be monitored, whereas the reliability is a value evaluated in a period before the timing to be monitored, and the two timings are different.

[Abnormal condition diagnosis method]
Hereinafter, an abnormal state diagnosis method by the above-described abnormal state diagnosis device 1 will be described with reference to FIGS. 2 to 5. The abnormal state diagnosis method according to the present embodiment creates M sub-models for predicting the state of a process from the actual values of a plurality of types of variables obtained during normal operation, and based on the prediction error of the sub-model, the process The deviation index from the normal state is calculated, and the cause of the abnormal state occurring in the process is estimated based on the deviation index pattern set including M deviation indexes calculated for each sub model. Here, the sub model does not necessarily have to be newly created, and if there is an existing model, the existing model can be used.

Specifically, as shown in FIG. 2, the abnormal state diagnosis method includes a reading step, a deviation index calculating step, an abnormality detecting step, a base pattern estimating step, and an abnormal cause estimating step. Further, in the abnormal state diagnosis method, a sub-model creating step, a base pattern creating step, and a base pattern learning step are performed as necessary. In the following description, an example in which the abnormal state diagnosis method is applied to a manufacturing process such as a steel process will be described.

<Reading step>
In the reading step, the deviation index calculating unit 71 reads the plurality of types of variable data acquired from the manufacturing process at the processing target time from the operation DB 31 (step S1).

<Deviation index calculation step>
Subsequently, in the deviation index calculation step, the deviation index calculation unit 71 uses the plural types of variable data read in the reading step to determine the manufacturing state of the manufacturing process at the processing target time as the manufacturing state of the manufacturing process during normal operation. A value indicating how different is is calculated as a deviation index for each sub-model (step S2).

Specifically, the deviation index calculation unit 71 first acquires the data of the sub-model from the sub-model DB 41, and substitutes the variable data read from the operation DB 31 into the corresponding sub-model to obtain the processing target time for each variable. Calculate the predicted value in. Next, the deviation index calculating unit 71 normalizes the data of the actual value and the predicted value of the plurality of types of variables in order to standardize the difference in absolute amount and unit between variables. Next, the deviation index calculation unit 71 uses the difference value between the normalized predicted value and the normalized actual value of the variable at the processing target time as the deviation index from the normal state of the manufacturing process for each sub model. calculate.

FIG. 3 shows the size of the deviation index for each sub-model when an abnormal state occurs in the manufacturing process. The vertical axis of the graph shows the size of the deviation index, and the horizontal axis shows the sub-model number. There is. In addition, in FIG. 1 to No. An example using 56 sub-models up to 56 is shown. Further, in the present embodiment, as shown in the figure, the size of the deviation index for each of the plurality of sub-models is collectively defined as a “deviation index pattern set”.

<Sub model creation step>
Here, in this embodiment, the sub-model creating step is performed before the timing of executing the deviation index calculating step. In the sub-model creation step, the sub-model creation unit 51 acquires a plurality of types of variable data obtained during normal operation from the operation DB 31, and uses M number of sub-models (main book) for predicting the manufacturing state of the manufacturing process from the variable data. In the embodiment, 56 pieces are created. Then, the sub model creating unit 51 stores the created M sub models in the sub model DB 41. Further, as described above, the sub model does not necessarily have to be created by the model creating unit 51, and if there is an existing model, it may be stored in the sub model DB 41 in advance without causing the model creating unit 51 to function. It is possible.

In this embodiment, it is assumed that the sub-model is configured by a regression model that models the correlation between 56 variables, as shown in the following formula (1), for example. In Expression (1) below, “XXX” is a variable corresponding to the sub model, and “aXX” is the weight of each variable.

<Abnormality detection step>
Subsequently, in the abnormality detection step, the abnormality detection unit 72 detects the abnormal state of the manufacturing process based on the deviation index calculated in the deviation index calculation step (step S3). The abnormality detection unit 72 detects that the manufacturing process is in an abnormal state, for example, when the deviation index calculated in the deviation index calculation step exceeds a predetermined threshold value.

Note that, for example, when considering the reliability of the sub-model for the deviation index, the abnormality detection unit 72 weights each deviation index according to the degree of reliability of the sub-model, and further sets the deviation index after weighting. It may be possible to detect that the manufacturing process is in an abnormal state when the total value exceeds a predetermined threshold value.

<Base pattern estimation step>
Subsequently, in the base pattern estimation step, the base pattern estimation unit 73 estimates the base pattern from the deviation index pattern set calculated in the deviation index calculation step (step S4).

Here, in the present embodiment, the base pattern creating step and the base pattern learning step are executed before the timing of executing the base pattern estimating step. Below, the contents of these steps will be described first.

<Basic pattern creation step>
In the base pattern creating step, the base pattern creating unit 52 creates a base pattern. FIG. 4 shows an example of the base pattern, in which the vertical axis of the graph shows the size of the deviation index and the horizontal axis shows the sub-model number. In the figure, only the bottom "basic pattern No. 56" shows the specific number of the sub-model on the horizontal axis of the graph, and the basic pattern Nos. 01, No. 02, No. 27, No. Illustration is omitted in 28 (the same applies to FIG. 5 described later). The base pattern is a pattern of anomalies shown when there is an anomaly in each of the variables related as explanatory variables of the sub model. The base pattern creation unit 52 creates M such base patterns for each explanatory variable that is an input signal.

Specifically, the base pattern creation unit 52 acquires, from the operation DB 31, normal state data of the manufacturing process necessary for calculating M (56 in the present embodiment) sub-models, and uses this as a reference data set. To do. Next, the base pattern creation unit 52 creates M perturbation data sets by adding one small abnormality to each of the variables related as explanatory variables of the sub-model to the reference data set.

For example, the base pattern creation unit 52 uses the sub model number. By artificially adding one small anomaly to each explanatory variable other than the explanatory variable “X1” for the item of the reference data set corresponding to the prediction target of 01 (“X1” in the above formula (1)) , Sub-model No. Create a perturbation data set for 01. In addition, the base pattern creation unit 52 uses the sub model number. By artificially adding one small abnormality to each explanatory variable other than the explanatory variable “X2” for the item of the reference data set corresponding to the prediction target of 02 (“X2” in the above formula (1)) , Sub-model No. Create a perturbation data set for 02. Similarly, the base pattern creation unit 52 creates a total of M×(M−1) (56×55 in this embodiment) perturbation data sets.

Next, the base pattern creation unit 52 creates a total of M (56 in this embodiment) deviation index pattern sets by supplying the created perturbation data set as an input of the sub-model. The perturbation data set created above can be classified into M types of data including micro abnormalities for each explanatory variable, and can be set as M data sets.

Specifically, the data of the model that predicts other than "X1" in which only a small abnormality is added to the explanatory variable "X1" is aggregated into one and given to the model other than X1. Next, the data of the model predicting other than "X2", which is a small abnormality added only to the explanatory variable "X2", is aggregated into one and given to the model other than X2. By repeating this, the base pattern is obtained. Note that it is assumed that there is no submodel that predicts the value of a specific explanatory variable, and in that case, there is no such submodel and it is excluded from the target of the base pattern (an explanatory variable of another submodel). Used only as). These are the base patterns shown in FIG. Then, the base pattern creating unit 52 stores the created base pattern in the base pattern DB 42.

When artificially adding a micro-abnormality to the above-mentioned reference data set, a certain value of the micro-abnormality may be added for each item corresponding to the prediction target of the sub-model. In the base pattern of the above, a small abnormality is added so that the maximum value of the deviation index is 10 times that in the normal state, that is, the maximum value of the deviation index in each of the base patterns is in a state that is normalized to some extent. Good.

<Basic pattern learning step>
Here, as shown in FIG. 28, the sub model No. Since a micro abnormality is given to the sub model No. 28, the sub model No. The deviation index of 28 shows a naturally high value, but other sub-model Nos. 13, No. The deviation index of 42 also shows a high value to some extent. That is, in reality, the sub model number. Even if 28 indicates a true abnormality, there are some other submodels that show an apparently high deviation index. Therefore, in the deviation index pattern set calculated from the actual operation data, it is not possible to judge that all deviation index patterns having a high deviation index are abnormal.

On the other hand, the deviation index pattern set calculated from the actual operation data can be considered to be a linear combination of the base patterns as shown in the following equation (2).

Therefore, if the basis pattern included in the deviation index pattern can be inversely estimated from the deviation index pattern calculated from the actual operation data, the true abnormality can be specified. However, in inverse estimation, it is usually difficult to uniquely determine the above-mentioned base pattern.

Therefore, in the base pattern learning step, the base pattern learning unit 61 creates a plurality of artificial deviation index pattern sets by combining a plurality of types of the base patterns created in the base pattern creating step, and the artificial deviation index pattern sets are included in the artificial deviation index pattern set. The types of underlying patterns that exist are learned by machine learning.

Specifically, the basic pattern learning unit 61 selects and acquires a plurality of basic patterns created in the basic pattern creating step from the basic pattern DB 42. Next, the base pattern learning unit 61 creates a plurality of artificial departure index pattern sets (hereinafter referred to as “artificial departure index pattern sets”) by adding and adding appropriate weights to each selected base pattern. To do. Next, the base pattern learning unit 61 performs machine learning using the artificial deviation index pattern set as an input and the type or weight of the selected base pattern as an output. Then, the base pattern learning unit 61 stores the learning result of the base pattern in the learning result DB 43.

Deep learning is effective as the machine learning performed in the base pattern learning step. Deep learning is a learning method that requires a multi-layered net structure, and is characterized by learning for each layer. In addition, deep learning has achieved high results in artificial intelligence in Go and image recognition. By using such a learning method, it is possible to stochastically estimate which base pattern is included in the deviation index pattern set.

Here, as the learning result of the base pattern, for example, there is information regarding the probability that a certain deviation index pattern set includes M (56 in the present embodiment) base patterns. In this case, in the learning result of the base pattern, the probability that the base pattern No. 01 is included is XX% and the probability that the base pattern No. 02 is included is XX% for a certain deviation index pattern set. ,..., “Probability that the base pattern No. 56 is included is XX%” is included.

Hereinafter, it returns to the specific description of the base pattern estimation step. In the base pattern estimation step, the base pattern included in the deviation index pattern set is estimated based on the learning result of the base pattern in the base pattern learning step described above. Specifically, the base pattern estimation unit 73 acquires the learning result of the base pattern from the learning result DB 43, and based on the learning result, calculates the deviation index pattern set obtained during actual operation, that is, the deviation index calculation step. The type of base pattern included in the set deviation index pattern set is estimated.

FIG. 5 shows an example of the result of estimating the types of base patterns included in the deviation index pattern set in the base pattern estimation step. As shown in the figure, the base pattern estimation unit 73 includes, for example, two base patterns (base patterns No. 30 and No. 24 in the figure) in the deviation index pattern set calculated in the deviation index calculation step. Presumed to be included. In addition, the base pattern estimation unit 73 estimates, for example, a predetermined number (two in the figure) of M base patterns that are 80% or more likely to be included in the deviation index pattern set. Enumerate as.

<Error cause estimation step>
Finally, in the abnormality cause estimation step, the abnormality cause estimation unit 74 estimates the cause of the abnormal state from the base pattern (step S5). The abnormality cause estimation unit 74 specifically estimates the cause of the abnormal state that has occurred in the manufacturing process, based on the type of the base pattern estimated in the base pattern estimation step and the combination of the base patterns.

For example, as shown in FIG. 5, the abnormality cause estimation unit 74 sets the deviation index pattern set to the base pattern No. in the base pattern estimation step. 30, No. If it is estimated that the base pattern No. 24 is included. 30, No. The cause of the abnormal state corresponding to the combination of 24 (for example, deterioration of the mechanical accuracy of a certain equipment) is estimated.

The correspondence between the type of base pattern or the combination of a plurality of base patterns and the cause of the abnormal state is, for example, experimentally obtained in advance and stored as a table. Then, the abnormality cause estimation unit 74 estimates the cause of the abnormal state corresponding to the type of the base pattern or the combination of the base patterns by referring to this table.

According to the abnormal state diagnosis method according to the present embodiment that performs the above processing, even when the same trouble has not occurred in the past, the cause of the abnormal state that occurred in the process such as the manufacturing process can be quickly identified. , Can be dealt with.

As described above, the abnormal state diagnosing method and the abnormal state diagnosing device of the process according to the present invention have been specifically described by the modes for carrying out the invention, but the gist of the present invention is not limited to these descriptions. It should be broadly construed based on the appended claims. It goes without saying that various changes and modifications based on these descriptions are also included in the spirit of the present invention.

For example, the abnormal state diagnosis device 1 can perform the abnormal state diagnosis method using the sub model, the base pattern, and the learning result of the base pattern that are created in advance and stored in each DB of the storage unit 40. In this case, the abnormal state diagnosis device 1 does not have to include the sub model creation unit 51, the base pattern creation unit 52, and the base pattern learning unit 61 shown in FIG.

Further, in the abnormal state diagnosis device 1, an example in which the sub model is configured by a regression model as shown in the above formula (1) is shown, but the specific configuration of the sub model is not particularly limited, and is configured by, for example, a physical model. You may.

Further, in the above-described embodiment, an example in which the abnormal state diagnostic device 1 and the abnormal state diagnostic method are applied to a manufacturing process such as a steel process has been described, but the abnormal state diagnostic device 1 and the abnormal state are used for a power generation process, a transportation process, and the like. It is also possible to apply diagnostic methods.

1 Abnormal State Diagnosis Device 10 Input Unit 20 Output Unit 30 External Device 31 Operation Database (Operation DB)
40 storage unit 41 sub-model database (sub-model DB)
42 Base Pattern Database (Base Pattern DB)
43 Learning Result Database (Learning Result DB)
50 model definition unit 51 sub-model creation unit 52 basis pattern creation unit 60 learning unit 61 base pattern learning unit 70 control unit 71 departure index calculation unit 72 anomaly detection unit 73 base pattern estimation unit 74 anomaly cause estimation unit

Claims (7)

A deviation index from the normal state of the process is calculated by using a plurality of submodels that predict the state of the process, and the process is generated based on the deviation index pattern set including the deviation index calculated for each submodel. A method of diagnosing an abnormal state of a process for estimating the cause of the abnormal state,
In the deviation index pattern set, for a variable related as an explanatory variable of the sub-model, a base pattern creation step of creating a pattern of anomalies shown when there is an anomaly as a plurality of base patterns,
An abnormal cause estimating step of estimating the cause of the abnormal state occurring in the process based on the base pattern;
A method for diagnosing an abnormal state of a process, comprising: A base pattern learning step of learning a type of the base pattern included in the artificial departure index pattern set by creating a plurality of artificial departure index pattern sets by combining a plurality of types of the base patterns created in the base pattern creating step When,
A basis pattern estimation step of estimating the type of the basis pattern included in the deviation index pattern set obtained during actual operation, based on the learning result in the basis pattern learning step,
Further including,
The method according to claim 1, wherein the abnormal cause estimating step estimates the cause of the abnormal state occurring in the process based on the type of the base pattern. In the base pattern creating step, one micro-abnormality is set for each variable related as an explanatory variable of the sub-model with respect to the reference data set including the normal state data of the process required for the calculation of the sub-model. 3. The process according to claim 1 or 2, characterized in that the perturbation data set is created by adding each one, and the basis pattern is created by giving the perturbation data set as an input of the sub-model. Abnormal condition diagnosis method. The base pattern learning step selects a plurality of the base patterns created in the base pattern creating step, adds a suitable weight to each of the selected base patterns, and adds the plurality of artificial deviation index pattern sets. 3. The method for diagnosing an abnormal state of a process according to claim 2 , wherein the machine learning is performed by using the artificial deviation index pattern set created as an input and the type or weight of the selected base pattern as an output. A deviation index from the normal state of the process is calculated by using a plurality of submodels that predict the state of the process, and the process is generated based on the deviation index pattern set including the deviation index calculated for each submodel. An abnormal condition diagnostic device for a process for estimating the cause of an abnormal condition,
In the deviation index pattern set, for a variable related as an explanatory variable of the sub-model, a base pattern creating unit that creates a pattern of anomalies shown when there is an anomaly as a plurality of base patterns,
Based on the basis pattern, an abnormal cause estimating means for estimating the cause of the abnormal state occurred in the process,
An abnormal condition diagnosis device for a process, comprising: A base pattern learning unit that creates a plurality of artificial departure index pattern sets by combining a plurality of types of the base patterns created by the base pattern creating unit and learns the types of the base patterns included in the artificial departure index pattern set When,
Based on a learning result in the base pattern learning means, a base pattern estimation means for estimating the type of the base pattern included in the deviation index pattern set obtained during actual operation,
Further equipped with,
The abnormal condition diagnosing device for a process according to claim 5, wherein the abnormal cause estimating means estimates the cause of the abnormal condition occurring in the process based on the type of the base pattern. The base pattern creating means has one micro-abnormality for each variable related as an explanatory variable of the sub-model with respect to a reference data set consisting of data of a normal state of the process necessary for calculation of the sub-model. 7. The process according to claim 5 or 6, wherein the basis pattern is created by creating a perturbation data set by adding the perturbation data sets as input to the sub-model. Abnormal condition diagnosis device.

Images