JP2020149259A - Abnormality diagnostic device - Google Patents

Abstract

To provide an abnormality diagnostic device in which abnormality contribution for determining an abnormal cause does not greatly change even when data do not greatly change.SOLUTION: An abnormality diagnostic device comprises a data classification unit which includes: a category calculation unit which classifies operation data output from a plurality of sensors provided in a diagnosis target into a plurality of categories; a reference point calculation unit which calculates a reference point of the plurality of categories; and an abnormality degree calculation unit which compares the reference point and current operation data, and calculates the degree of abnormality of the current operation data. The reference point is a weighted average of two or more categories out of the plurality of categories.SELECTED DRAWING: Figure 7

Description

The present invention relates to an abnormality diagnostic device of a plant or a device.

Power plants and related equipment are equipped with many sensors such as thermometers, pressure gauges and flow meters for monitoring and control purposes. In recent years, sensors such as acceleration sensors have been installed in infrastructure equipment such as bridges and roads to monitor the condition of the equipment.

The feature of such sensor data is that there are a plurality of related sensor data and it is a multidimensional time series data measured in a certain time cycle. Various methods have been proposed as condition monitoring methods that utilize such multidimensional time series data.

For example, in Patent Document 1, as a technique using adaptive resonance theory, "it belongs to a category in which a plurality of measurement data from various sensors installed in a plant to be diagnosed are discriminated from normal data by adaptive resonance theory. A plant abnormality diagnosis device including an abnormality degree calculation unit for obtaining an abnormality degree of the entire plant to be diagnosed based on a difference in spatial distance between the data and the plurality of measurement data is disclosed. ..

Here, the adaptive resonance theory (ART: Adaptive Resonance Theory, hereinafter referred to as ART) is a method for classifying multidimensional data into a plurality of categories according to the degree of similarity, and is described in, for example, Non-Patent Document 1. .. By using ART, it is possible to detect anomalies by clustering multidimensional data and dividing it into categories.

International Publication No. 2018/052568

G.A.Carpenter and S.Grossberg, “ART2: Self-OrganizatioNo.f stable category recognition codes for analog input patterns”, Applied Optics, Vol26, No.23, 1987

In Patent Document 1, as shown in FIG. 1, an abnormality in equipment is detected based on operation data (diagnosis data) in which normal operation data (learning data) deviates from the classified category (normal category). In order to determine the degree of abnormality and which signal deviation is large, the degree of abnormality and the degree of abnormality contribution are calculated based on the difference between the normal data and the current measurement data.

Here, the degree of abnormality and the degree of contribution to abnormality in Patent Document 1 will be described with reference to the drawings.

As shown in FIG. 2, when an abnormality is detected, the distance from the current data k to the center of gravity of the closest normal category (category 3 in the example of FIG. 2) is defined as the degree of abnormality. The degree of anomaly indicates the degree of abnormality in the data.

Further, as shown in FIG. 3, the anomaly degree vector decomposed into the components of each measured value (x1 and x2 in the example of FIG. 3) is defined as the anomaly contribution degree. The anomalous contribution indicates which signal has a large deviation from normal.

However, when the degree of anomaly and the degree of anomaly contribution described in Patent Document 1 are used, there are cases where the degree of anomaly contribution changes significantly even though the measurement data at the time of abnormality does not change significantly. An example thereof will be described with reference to FIG.

FIG. 4 shows a situation in which new data m is measured in the vicinity in addition to the state in which the data k shown in FIGS. 2 and 3 is abnormal. The normal category closest to the data k is category 3, but the normal category closest to the data m is category 1. As a result, the values of the degree of anomaly of both data do not change significantly, but the degree of anomaly contribution changes significantly. That is, in the data k, the abnormal contribution of x2 was large, whereas in the data m, the abnormal contribution of the data x1 was large.

As described above, in the method described in Patent Document 1, although the measurement data itself has not changed significantly, the anomalous contribution may change significantly and the cause of the anomaly cannot be estimated based on the anomaly contribution. It was.

In order to solve the above problems, the abnormality diagnosis device according to the present invention has "a category calculation unit that classifies the operation data output from a plurality of sensors provided in the diagnosis target into a plurality of categories, and a reference of the plurality of categories. It is provided with a data classification unit having a reference point calculation unit for calculating points, an abnormality degree calculation unit for comparing the reference point with the current operation data, and calculating the abnormality degree of the current operation data, and the reference point is , It is a weighted average of two or more categories out of a plurality of categories. "

Since the present invention has the above configuration, the anomaly contribution degree does not change significantly in the measurement data in the vicinity, and the accuracy of estimating the cause of the anomaly can be improved.

The figure which shows the method of classifying the conventional data. The figure which shows the example which calculated the degree of abnormality by the conventional method. The figure which shows the example which calculated the anomalous contribution degree by a conventional method. The figure which shows the problem when the anomalous contribution degree is calculated by the conventional method. The figure which shows the structure of the Example of this invention. The figure which shows the example of the operation data stored in the operation data database. The figure which shows the example of the structure of the data classification part of the Example of this invention. The figure which shows the calculation method of the degree of abnormality of this invention. The figure which shows an example of the trend graph of the classified category. The figure which shows an example of the trend graph of the degree of anomaly and the degree of anomaly contribution. The figure which shows the test data. The figure which shows the example which calculated the degree of abnormality by the conventional method using the test data. The figure which shows the example which calculated the anomaly contribution degree by a conventional method using test data. The figure which shows the example which calculated the degree of abnormality by the method of this invention using the test data. The figure which shows the example which calculated the anomalous contribution degree by the method of this invention using the test data.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. It should be noted that the following is merely an example of implementation, and the invention itself is not intended to be limited to the following specific contents.

FIG. 5 shows a configuration according to an embodiment of the present invention. This embodiment is an example in which the diagnostic apparatus of the present invention is used for abnormal diagnosis of a plant, and includes a plant 10, an operation data database 20, a data classification unit 30, a classification result database 40, and a display / operation unit 50. The outline of each component is described below. The target of abnormality diagnosis is not limited to plants, but can be applied to other devices and the like.

The plant 10 includes sensors and is used for plant control and monitoring. Examples of sensors include, but are not limited to, flow meters, thermometers, pressure gauges, and the like.

In the operation data database 20, the operation data of the plant measured by the sensor of the plant 10 is stored as time-series data every minute, for example. Of the stored operation data, the operation data designated as normal data by the display / operation unit 50 is cut out as learning data and sent to the data classification unit 30. In addition, the operation data database 20 also temporarily stores the data measured in real time by the plant 10, and then sends it to the data classification unit 30 as diagnostic data at regular intervals.

The data classification unit 30 classifies multidimensional operation data into a plurality of categories by using an ART-based clustering algorithm (hereinafter referred to as improved ART).

Here, ART will be described. In ART, the input data is classified into multiple categories (clusters), and the classified category numbers are assigned to each input data. A category represents a group of data having similarities, and input data to which the same category number is assigned has a high degree of similarity.

In the learning phase, the operation data (learning data) in the normal state of the equipment is input to ART. Since ART classifies driving data (learning data) into multiple categories according to the similarity of data, it defines the category (normal category) to be generated when the driving data is normal.

In the diagnosis phase, the operation data (diagnosis data) to be diagnosed is input to the ART that has learned the normal data. As a result, the data having a high degree of similarity to the learning data is classified into the same category as the learning phase. However, when the tendency of the data changes, such as when some abnormality occurs in the equipment, it is classified into a category (new category) different from the training data.

The data classification device 30 using the improved ART according to the present embodiment outputs the category number, the degree of abnormality, and the degree of abnormality contribution of each measurement data in which the input multidimensional data is classified as the data classification result.

The classification result database 40 manages the category number, the degree of abnormality, and the degree of contribution to abnormality output from the data classification unit 30. In addition, the weighting coefficient that is the basis for selecting each category and the centroid data of the data classified into each category are saved.

The display / operation unit 50 sets conditions for learning data and diagnostic data. Based on this condition, it is determined whether the operation data measured by the sensor of the plant 10 is learning data or diagnostic data. In addition, the display / operation unit 50 displays a trend graph of the category number, the degree of abnormality, and the degree of abnormality contribution. The display / operation unit 50 may be configured such that the display unit and the operation unit are separated from each other.

Next, the method of abnormality diagnosis according to this embodiment will be described in detail with reference to FIGS. 6 to 10.

The plant 10 is composed of equipment, pipes connecting the equipment, valves, and the like. Sensors such as flowmeters, thermometers, and pressure gauges are installed in the equipment and piping to monitor and control the condition of the plant. For example, in the case of a thermometer, these sensors are tagged with "TIC001". This tag becomes the ID of each sensor, and each sensor is identified by the ID.

The operation data database 20 records the data measured by the sensors of the plant 10 as time series data. An example of operation data is shown in FIG. As shown in FIG. 6, as the operation data for each hour measured by the sensor installed in the plant 10, the time when the operation data was measured is in the "Time" column, and "FIC001" and "PIC001" are in the other columns. The value measured by the sensor specified by the ID such as is recorded. This time interval can be arbitrarily specified, but in this embodiment, it is set to 1 minute.

The data classification unit 30 classifies the operation data using the improved ART. The detailed configuration of the data classification unit 30 is shown in FIG. The data classification unit 30 includes a category calculation unit 31, a reference value calculation unit 32, an abnormality degree calculation unit 33, and an abnormality contribution calculation unit 34.

The category calculation unit 31 classifies the driving data into categories according to the similarity of the driving data. The category in which driving data (learning data) is classified in the learning phase is regarded as a normal category. Since detailed algorithms for classifying into categories are described in Non-Patent Document 1 and Patent Document 1, description thereof will be omitted here.

The reference point calculation unit 32 calculates a reference point for calculating the degree of abnormality. The degree of abnormality is calculated based on the deviation information between the reference point and the current operation data (diagnostic data).

A method of calculating the degree of abnormality according to the present invention will be described with reference to FIG. In the present embodiment, it is calculated by equation (1) the weighted average g _a of the center of gravity of the normal range as a reference point.

Here, g _i is the normal categories j of the centroid vector (coordinates), d _i ^(k) is the distance from g _i to the current measurement data k. f (x) is a monotonous decrease function of x, and in this embodiment, the sigmoid function of Eq. (2) was used as an example.

According to equation (2), the distance in accordance with the d _i ^(k) is increased, f (d _i ^(k)) becomes smaller. That is, the coefficient of the weighted average ga _a is large in the normal category close to the measurement data m, and is small in the far category. In particular, since the sigmoid function used in Eq. (2) is a function whose value gradually approaches zero as x increases, the coefficient of the category with a large distance d _i ^(k) approaches zero, which affects the calculation of the reference point. Do not give.

That is, the reference point obtained by the equation (1) is the average value of the center of gravity of the category close to the measurement data among the coordinates of the center of gravity of each category.

The abnormality degree calculation unit 33 obtains the abnormality degree of the current measurement data k from the coordinates of the reference point obtained by the reference point calculation unit 32 and the coordinates of the current measurement data k. The degree of anomaly is defined by the distance between the two, and the magnitude of the degree of anomaly vector d _a ^(k) in Eq. (3) is the degree of anomaly.

The anomaly contribution calculation unit 34 obtains the anomaly contribution R _i of the parameter _i by the following equation (4), where each component of the anomaly vector d _a ^(k) obtained by the anomaly calculation unit 33 is x _i .

The classification result database 40 stores the category number, abnormality degree, and abnormality contribution degree of each data obtained by the data classification unit 30.

The display / operation unit 50 displays the data stored in the classification result database 40 and the operation data stored in the operation data database 20. The trend graph of the category classified in FIG. 9 is shown. It can be seen that the category changes with the change of time and the state of the plant changes. In the example of FIG. 9, categories 1 to 3 are normal categories, and category 4 is a new category.

FIG. 10 shows a display example of the degree of abnormality and the degree of contribution to abnormality at the same time as in FIG. What is shown by the black line is the degree of abnormality. Since the anomalous contribution is calculated by the equation (4), the sum of the anomalous contributions of each parameter is the anomaly. The example shown in the figure, the parameter is two examples, an abnormal contribution R ₁ and parameter x ₂ abnormality contribution R ₂ a total error probability parameters x _1. Comparing FIGS. 9 and 10, it can be seen that the degree of abnormality increases from the time when the category becomes category 4, which is a new category.

Next, the results of comparing the conventional algorithm and the algorithm shown in this embodiment using test data will be described with reference to FIGS. 11 to 15.

FIG. 11 shows two-dimensional test data. There were 50 training data, and as a result of classification by ART, they were classified into 9 categories. The points shown in ◆ indicate the center of gravity of each category. There are 31 points of diagnostic data, and at the time of diagnosis, the data were input in order from left to right.

The results of determining the degree of anomaly and the degree of anomaly contribution by the conventional algorithm are shown in FIGS. 12 and 13. The anomaly degree shown in FIG. 12 does not change smoothly at 7 minutes and 21 minutes, but there is no discontinuity, and there is no major problem in calculating the anomaly degree even with the conventional algorithm. You can see that. On the other hand, in the anomalous contribution shown in FIG. 13, the anomalous contribution of the center of gravity of the closest normal category changes discontinuously at 7 minutes. If the anomalous contribution changes discontinuously due to changes in the closest normal category, it may be misjudged to estimate the cause of the anomaly using this data.

Next, the results of determining the degree of anomaly and the degree of anomaly contribution by the algorithm of the present invention are shown in FIGS. 14 and 15. Compared with the case of using the conventional algorithm, the tendency of the degree of anomaly does not change much, but it can be seen that there is no discontinuous change in the graph of the degree of anomaly contribution.

As described above, by using the abnormality diagnosis device of the present invention, the phenomenon in which the degree of abnormality contribution changes significantly even though the measurement data are close to each other is eliminated. That is, the accuracy when estimating the cause of an abnormality using the degree of anomaly contribution is improved.

In this example, ART is used as the data clustering technique, but other clustering techniques may be used. Moreover, although the sigmoid function was used as f (x), another function such as 1 / x may be used.

10: Plant 20: Operation data database 30: Data classification unit 31: Category calculation unit 32: Reference value calculation unit 33: Abnormality calculation unit 34: And abnormality contribution calculation unit 40: Classification result database 50: Display unit

Claims (10)

A category calculation unit that classifies the operation data output from multiple sensors provided in the diagnosis target into multiple categories, and
The reference point calculation unit that calculates the reference points of the plurality of categories, and
An abnormality degree calculation unit that compares the reference point with the current operation data and calculates the abnormality degree of the current operation data.
Equipped with a data classification unit
The abnormality diagnostic apparatus, wherein the reference point is a weighted average of two or more categories among the plurality of categories. The abnormality diagnosis device according to claim 1, wherein the abnormality degree calculation unit calculates an abnormality degree of the current operation data based on deviation information between the reference point and the current operation data. The abnormality diagnosis device according to claim 1 or 2, wherein the abnormality degree calculation unit calculates the abnormality degree of the current operation data based on the difference in the distance between the reference point and the current operation data. .. The abnormality diagnosis device according to any one of claims 1 to 3, wherein the operation data is multidimensional data composed of measurement data output from each of the plurality of sensors. The abnormality diagnosis device according to claim 4, further comprising an abnormality contribution calculation unit that calculates a component of the abnormality degree in the axial direction for each of the plurality of measurement data as an abnormality contribution degree. It is provided with an operation data database that outputs the operation data when the diagnosis target is normal among the input operation data to the data classification unit as learning data.
The category calculation unit classifies the category in which the learning data is classified as a normal category.
The abnormality diagnostic apparatus according to any one of claims 1 to 5, wherein the reference point is a weighted average of two or more of the normal categories. The abnormality diagnostic apparatus according to any one of claims 1 to 6, wherein a weighted average coefficient is defined based on a distance from the current data to the reference point. The abnormality diagnosis apparatus according to claim 7, wherein the coefficient of the weighted average becomes smaller as the distance from the current data to the reference point becomes larger. The abnormality diagnosis device according to any one of claims 1 to 8, further comprising a display / operation unit for displaying the classification result output from the data classification unit. The abnormality diagnosis device according to any one of claims 1 to 9, wherein the classification result includes a category number classified by the category calculation unit and the degree of abnormality.

Images