WO2014091952A1

WO2014091952A1 - Sensor monitoring device, sensor monitoring method, and sensor monitoring program

Info

Publication number: WO2014091952A1
Application number: PCT/JP2013/082303
Authority: WO
Inventors: 哲寺澤; 加藤　真也; 敬之山本; 敬喜朝倉; 林　司; 山本　秀夫; 睦男生田; 安達　勝; 将弘崎部; 健三宮; 知也相馬; 真弓高城; 大石　敏之
Original assignee: 日本電気株式会社; 中国電力株式会社
Priority date: 2012-12-14
Filing date: 2013-11-25
Publication date: 2014-06-19
Also published as: JPWO2014091952A1; JP6304767B2

Abstract

Provided is a technology that monitors sensors themselves and prevents misdiagnosis caused by deterioration over time and environmental changes. This sensor monitoring device provided in power plants comprises: a plurality of sensor sets each including a plurality of sensors; a database that models the correlation between sensor values from the plurality of sensors configuring each sensor set and the sensor values from a plurality of sensors configuring other sensor sets, and stores same as a relationship model; and a correlation detection module. The correlation detection module: compares the correlation between the sensor detection values for the plurality of sensors configuring each sensor set and the sensor detection values for the plurality of sensors configuring the other sensor sets, and the correlation to the modeled relationship model stored in the database, and detects deviation from the correlation to the modeled relationship model, in the correlation between the detected sensor values.

Description

SENSOR MONITORING DEVICE, SENSOR MONITORING METHOD, AND SENSOR MONITORING PROGRAM

The present invention relates to a sensor monitoring device, a sensor monitoring method, and a sensor monitoring program for monitoring the state of a sensor, and more particularly to a sensor monitoring device, a sensor monitoring method, and a sensor monitoring program for monitoring a sensor installed in a power plant.

At power plants, various devices are diagnosed and maintenance work is performed when necessary for the purpose of operating safely and efficiently. From the viewpoint of safety, especially in power plants such as nuclear power plants, many sensors are installed, and various types of parameters are constantly measured and diagnosed.
Diagnosis of various devices is performed by a plurality of sensors attached to a device such as a water supply pump. Usually, it is understood that some abnormality has occurred in the device when the measured value detected by the sensor exceeds a predetermined threshold. Here, when sensor drift occurs in the sensor due to aging deterioration or environmental change, the measured value deviates from the correct value. That is, when any abnormality is recognized in the measurement value detected by the sensor, it cannot be determined whether an abnormality has occurred in the device or in the sensor itself. Therefore, in order to determine whether the device should be maintained or the sensor should be maintained, it is necessary to distinguish between a device abnormality and a sensor drift.
In addition, when the measured value measured by the sensor reaches a predetermined threshold value, it is determined that there is an abnormality. However, if the measured value changes over a long period of time, such as sensor aging, There is a possibility that the discovery will be delayed and an accurate diagnosis cannot be made, leading to the shutdown of the power plant.
In Patent Document 1, various process quantities are measured in a plant such as a nuclear power plant, and the possibility that the process value may be abnormal is added to the data as ambiguity to form process data. In consideration of the possibility of failure, a technique for providing a plant diagnosis system that accurately detects an abnormality in a process value is disclosed.

Japanese Patent Application Laid-Open No. 06-128983

However, in the diagnosis of various devices, there is a problem that it is difficult to distinguish between a change in the process value due to sensor drift and a change in the process value itself. In the related technology, a technique for removing the drift amount of the sensor investigated in advance from the process value used for diagnosis of various devices has been adopted. In this method, if an incorrect drift estimation value is input when the drift is removed from the process value measured by the sensor, there is a possibility of erroneous diagnosis.
An object of the present invention is to provide a sensor monitoring apparatus, a sensor monitoring method, and a sensor monitoring program that solve the above-described misdiagnosis.

In view of the above-described problems, one embodiment of the present invention is a sensor monitoring device provided in a power plant, which includes a plurality of sensor sets each including a plurality of sensors, and sensors from the plurality of sensors constituting each sensor set. A database for modeling correlation between values and sensor values from the plurality of sensors constituting another sensor set, and storing them as a relationship model, and sensor detection values in the plurality of sensors constituting each sensor set And the correlation between the sensor detection values of the plurality of sensors constituting the other sensor set and the correlation of the modeled relationship model stored in the database and comparing the detected sensor values. A correlation detection module for detecting disturbances from the correlation of the modeled relationship model in the correlation of It relates to a sensor monitoring apparatus characterized by.
Another aspect of the present invention is a sensor monitoring method executed at a power plant, wherein sensor values from a plurality of sensors constituting each sensor set and a plurality of sensors constituting another sensor set are received. Referring to a database storing a relationship model in which a correlation with a sensor value is modeled, sensor detection values in the plurality of sensors constituting each sensor set and sensors in the plurality of sensors constituting another sensor set The correlation between the detected values is compared with the correlation of the modeled relationship model stored in the database and the correlation of the modeled relationship model in the correlation of the detected sensor value The present invention relates to a sensor monitoring method characterized by detecting disturbance from a relationship.
Furthermore, another aspect of the present invention is a sensor monitoring program executed in a power plant, wherein sensor values from a plurality of sensors constituting each sensor set and a plurality of sensors constituting another sensor set are received. A process for referring to a database storing a relationship model in which a correlation with a sensor value is modeled, a sensor detection value in the plurality of sensors constituting each of the sensor sets, and the plurality of sensors constituting another sensor set The process of comparing the correlation between the sensor detection values in FIG. And the correlation of the modeled relationship model stored in the database and the modeled relation in the correlation of the detected sensor values The present invention relates to a sensor monitoring program for causing a computer to execute processing for detecting disturbances from the correlation of models.

According to the present invention, it is possible to identify an anomaly that has occurred due to a disorder in the correlation and discover it at an early stage, thereby improving the safety of the power plant.
Further advantages and embodiments of the present invention are described in detail below using the description and the drawings.

FIG. 1 is a block diagram showing a schematic configuration of a system according to an embodiment of the present invention.
FIG. 2 is a flowchart for explaining a model generation phase for constructing a relationship model executed in one embodiment of the present invention.
FIG. 3 is a flowchart for explaining a monitoring phase executed in an embodiment of the present invention.
FIG. 4 is a graph showing sensor measurement values of the respective sensors installed at the same location in the related art.

First, before describing the contents of the present invention, the technology studied by the present inventors and the problems thereof will be described. Currently, when a correct measurement value is determined from sensor detection values measured by a plurality of sensors, a determination based on an empirical rule is made. A method for determining sensor measurement values based on this rule of thumb will be described with reference to FIG.
FIG. 4 is a graph showing sensor measurement values (detection values) of a plurality of sensors arranged in the same place with time. In this example, the sensor measured values by the three sensors # 1, # 2, and #N are graphed. As can be seen from FIG. 4, when a process value is measured by a plurality of sensors installed at the same physical location, if the measured value of each of the plurality of sensors is the same, the measured value may be used as a correct value. it can. The problem here is when the measured values of the plurality of sensors are different. This is because it is unknown which sensor's measured value is correct. For example, in the present situation, an intermediate value (measured value of sensor # 2) is adopted, or an appropriate rule of measurement is determined by predicting an appropriate sensor measurement value by judging from an empirical rule which sensor becomes abnormal due to deterioration. It has been done. In addition, in this embodiment, a process value shows the state quantity showing the state of a power plant, and the detected value measured by the sensor may correspond directly to a process value. Further, it may be obtained by calculating a plurality of detection values.
However, in the method of determining the correct sensor measurement value by these methods, the selected measurement value is not always correct only by selecting the measurement value considered to be more correct. Therefore, if the power plant is operated using the measurement values selected in this way, there is a risk of making erroneous judgments with incorrect measurement values. Can cause serious problems.
An embodiment of the present invention that can solve the above problems will be described with reference to the drawings. However, the technical scope of the present invention is not construed as being limited by the embodiments described below.
FIG. 1 is a diagram showing a schematic configuration of a system (here, a power plant system) to which a sensor monitoring apparatus according to an embodiment of the present invention is applied.
The illustrated power plant system is composed of various devices, and power generation using nuclear power or thermal power is performed. In the present embodiment, various devices of the power plant system are the measurement target 11. Examples of the measurement object 11 include devices such as a nuclear reactor containment vessel, a pressurizer, a condenser, power wiring, a boiler, a turbine, and a water supply pump. Sensors S <b> 1 to S <b> 3 and other sensor groups 15 are appropriately arranged on these various devices to be measured 11.
The plurality of sensors S1 to S3 are attached to various devices of the system as a sensor set, and measure the process values of the system. In addition, one sensor set is physically installed at the same cylinder, and is arranged at a physical position different from that of the other sensor group (a plurality of sensor sets) 15. In a power plant, measured values are constantly measured by a large number of sensor sets composed of sensors S1 to S3 multiplexed in order to grasp the state, and are used for actual driving and safety decision making. Here, examples of the sensors S1 to S3 include sensors such as a temperature sensor, a voltage / current sensor, a vibration sensor, a radiation dose sensor, a pressure sensor, a flow rate sensor, and a water level sensor. Also, the number of sensors is not limited to the three illustrated in FIG. 1, and it goes without saying that four or five or more sensors can be installed for safety measures. The same applies to the other sensor groups 15. As will be described later, in the technique used in the present embodiment, the effectiveness of analysis increases as the number of sensors increases. The sensor set described above constitutes a part of the sensor monitoring apparatus according to the present embodiment.
The monitoring module 13 periodically acquires sensor detection values (measurement values) output from the plurality of sensors S1 to S3 installed in the system. The monitoring module 13 stores sensor detection values from the sensors S1 to S3 as sensor information in the sensor value storage database 17 together with IDs, times, physical positions, measurement target information, and the like of the sensors S1 to S3. Similarly, the monitoring module 13 periodically acquires sensor detection values from other sensor groups 15 and similarly stores them in the sensor value storage database 17.
Next, the data stored in the sensor value storage database 17 is input to the relationship verification module 19 (correlation detection module). Here, the relationship verification module 19 verifies the relationship between the sensor data of the plurality of sensor sets including the plurality of sensors S1 to S3 and the relationship between the sensor data of the plurality of sensors S1 to S3. That is, the relationship verification module 19 includes a relationship storage database 21. The relationship storage database 21 includes a plurality of sensors S1 to S1 as well as correlations between sensor data of the plurality of sensors S1 to S3. Relationships between past sensors and sensor sets that represent correlations between sensor data of a plurality of sensor sets including S3 are accumulated. As the relationship between past sensors and sensor sets, it is preferable to use output data of the sensors S1 to S3 and other sensor groups 15 when the system is operating normally.
The verification result in the relationship verification module 19 is given to the drift determination module (determination module) 23 to determine sensor drift. If an abnormality is detected as a result of the determination, an alarm or a display device notifies the user.
For convenience of explanation, in FIG. 1, the monitoring module 13, the relationship verification module 19, and the drift determination module 23 are illustrated as hardware, but these modules are actually configured by a program that can be executed by a computer. Can be done. In this case, the monitoring module 13, the relationship verification module 19, and the drift determination module 23 are realized by a storage medium that stores programs for executing operations in these modules and a computer (CPU) that executes these programs. it can.
As described above, according to the embodiment of the present invention, the correlation between the sensor detection values in the plurality of sensors constituting each sensor set and the sensor detection values in the plurality of sensors constituting another sensor set is performed by the relationship verification module 19. The relationship and the correlation of the modeled relationship model stored in the relationship storage database 21 are compared, and the disturbance from the correlation of the modeled relationship model in the correlation of the detected sensor values is compared. By detecting it, it is characterized by detecting an abnormality that has occurred at an early stage.
The above points will be described more specifically with reference to FIGS.
First, referring to FIG. 2, the generation phase of the relationship model stored in the relationship storage database 21 is shown. Here, the relationship model generation phase is described as being executed by a program that can be executed by a computer, but it can also be realized by using a hardware circuit. As shown in step F1-1 in FIG. 2, a sensor data collection period is defined and a collection period is set (F1-2). When the collection period is set, sensor data is acquired during the collection period from the sensors S1 to S3 and the other sensor groups 15 in step F1-3. The acquired sensor data is stored in a memory (not shown) included in the relationship verification module 19.
Subsequently, in step F1-4, the correlation between the sensor data from the plurality of sensors S1 to S3 and other sensor groups 15 is expressed under the control of the verification program for realizing the operation of the relationship verification module 19. A relationship model is built. The constructed relationship model is stored in the relationship storage database 21 in step F1-5, and the creation of the relationship model is terminated.
Next, a monitoring phase in which monitoring is performed using the generated relationship model will be described with reference to FIG. The monitoring phase shown in FIG. 3 is performed using the monitoring module 13, the sensor value storage database 17, the relationship verification module 19, the relationship storage database 21, and the drift determination module 23 shown in FIG. In practice, the monitoring phase is executed by a program that realizes an operation corresponding to the operation performed by the above-described module, but can also be realized by using a hardware circuit.
In the example of FIG. 3, the case where the present invention is applied to monitoring of the sensors S1 to S3 will be described. The monitoring phase shown in FIG. 3 is started by acquiring sensor detection values (detection data) from the sensors S1 to S3 in the monitoring module 13 (F2-1), and a sensor set including the sensors S1 to S3; Correlation with other sensor sets is confirmed in step F2-2. This operation is performed using the relationship verification module 19 and the relationship storage database 21.
In step F2-3, the relationship verification module 19 checks whether there is any disturbance in the correlation between the sensor data from each of the sensors S1 to S3 and the other sensor group 15 by correlation analysis. That is, the correlation model constructed using the sensor values obtained from the plurality of sensor sets including the plurality of sensors S1 to S3 and the sensor detection value from the sensor set including the plurality of sensors S1 to S3 are the same correlation. Is determined in step F2-3. Here, correlation analysis using a relationship model automatically generates an invariant model of the entire system (for example, a power plant) from a plurality of observation data (for example, sensor values from a sensor) obtained from a physical system. This refers to monitoring the state of the system (or the sensor itself) by comparing this model with data obtained from an actual physical system (for example, a sensor detection value in the sensor). When the correlation is not disturbed (step F2-3: NO), the process returns to step F2-1 and sensor detection values from other sensors are acquired.
On the other hand, if the correlation is disturbed in step F2-3 (step F2-3: YES), the relationship verification module 19 determines that an abnormality has occurred in the sensor set including the sensors S1 to S3. The process proceeds to step F2-4. In step F2-4, the relationship between the sensors S1 to S3 measuring the same location is checked. First, in step F2-5, the correlation of the sensor S1 with the other sensors S2 and S3 is compared and the value of the sensor S1 with a small disturbance is stored. This operation is performed for the number of sensors installed at the same location (step). In this example, since the verified sensor is the sensor S1 (step F2-7: NO), the process returns to steps 2-4 and F2-5, and the sensor S2 is similarly verified. Then, the correlation disturbance with the other sensors S1 and S2 in the last sensor S3 installed in the same place is compared, and a sensor with a small disturbance is stored (F2-5). After the sensor S3 that measures the same location is verified (F2-7: YES), the process proceeds to step F2-8, and the value of the sensor with the smallest disturbance among the verified sensors S1 to S3 is applied. The process ends. The reason why the value of the sensor having the smallest disturbance among the verified sensors S1 to S3 is applied is that the value of the sensor having the smallest disturbance is the most detected value of the sensors S1 to S3 installed at the same location. This is because the reliability is high.
As described above, in the present embodiment, by monitoring the disorder of the correlation between the multiplexed sensors and the collapse of the correlation between the sensor sets, the measured process value is valid or the sensor drift is detected. It is determined whether. By determining whether or not there is an effect due to sensor drift, if the effect is due to sensor drift, the measured value is corrected using the drift estimation value, which is a related technology, and it is possible to shift to inspection based on state standards instead of periodic inspection It becomes.
In the technology using the correlation analysis based on the relationship model used in the present embodiment, the more the number of sensors installed, the more effective analysis can be performed, so in an existing power plant where a large number of sensors are arranged. A very effective analysis can be performed. In addition, the present invention is particularly effective for grasping pressure abnormalities that affect a wide area in space and the situation by sensors that are spatially densely installed.
According to the present embodiment, an unhealthy state before an abnormality occurs, that is, a sign of a failure, is detected at an early stage, rather than a method of detecting an abnormality for the first time when a threshold is exceeded as in the current abnormality detection means. be able to. Further, in the present embodiment, since abnormality is diagnosed based on the disorder of correlation between the sensor to be audited and thousands of other sensors, it is possible to easily distinguish between abnormality of the device and sensor drift. .
In addition, the sensor monitoring program using the characteristics included in the above-described embodiment is also included in the category of the present invention. In the above-described embodiments, the processing of the embodiments may be executed by a computer-readable storage medium encoded with a program, software, or instructions that can be executed by a computer. The storage medium includes not only a portable recording medium such as an optical disk, a floppy (registered trademark) disk, and a hard disk, but also a transmission medium that temporarily records and holds data such as a network.
In the embodiment described above, for convenience of explanation, the database is divided into two databases, the sensor value storage database 17 and the relationship storage database 21. Needless to say, the sensor monitoring device, the sensor monitoring method, and the sensor monitoring program of the present invention can be configured by using the sensor value storage database 17 and the relationship storage database 21 as a single database.
The sensor monitoring apparatus, the sensor monitoring method, and the sensor monitoring program described in the above-described embodiments can be arbitrarily applied to a power plant such as a nuclear power plant or a thermal power plant.
The present invention has been described above with reference to the embodiments, but the present invention is not limited to the above embodiments. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.
The present invention can be implemented in various other forms without departing from the spirit or main features thereof. Therefore, the above-mentioned embodiment is only a mere illustration in all points, and should not be interpreted limitedly. The scope of the present invention is indicated by the scope of the claims, and is not restricted by the text of the specification. Furthermore, all modifications, various improvements, substitutions and modifications belonging to the equivalent scope of the claims are all within the scope of the present invention.
This application claims its benefit on the basis of priority from Japanese Patent Application No. 2012-273453 filed on Dec. 14, 2012, the disclosure of which is hereby incorporated by reference in its entirety. take in.

11 Measurement objects S1 to S3 Sensor 13 Monitoring module 15 Other sensor group 17 Sensor value storage database 19 Relationship verification module 21 Relationship storage database 23 Drift determination module

Claims

A sensor monitoring device provided in a power plant,
A plurality of sensor sets each including a plurality of sensors;
A database for modeling correlation between sensor values from the plurality of sensors constituting each sensor set and sensor values from the plurality of sensors constituting another sensor set, and storing as a relationship model;
Correlation between sensor detection values in the plurality of sensors constituting each sensor set and sensor detection values in the plurality of sensors constituting another sensor set, and the modeled relationship stored in the database A correlation detection module that compares the correlation of the model and detects a disturbance from the correlation of the modeled relationship model in the correlation of the detected sensor value. Monitoring device.
The sensor monitoring apparatus according to claim 1, further comprising a determination module that selects a sensor with less disorder in the correlation from the plurality of sensor sets.
The sensor monitoring apparatus according to claim 1 or 2, wherein the plurality of sensor sets are attached to different physical positions.
The sensor monitoring device according to claim 1 or 2, wherein the plurality of sensor sets are attached to different physical positions of the power plant.
5. The determination module according to claim 2, wherein the determination module estimates a drift amount in each sensor of the plurality of sensor sets, and performs correction based on the estimated drift amount. Sensor monitoring device.
A sensor monitoring method executed in a power plant,
Refer to a database that stores a relationship model that models the correlation between sensor values from multiple sensors that make up each sensor set and sensor values from multiple sensors that make up another sensor set,
Correlation between sensor detection values in the plurality of sensors constituting each sensor set and sensor detection values in the plurality of sensors constituting another sensor set, and the modeled relationship stored in the database Compare with model correlation,
A sensor monitoring method, comprising: detecting a disturbance from the correlation of the modeled relationship model in the correlation of the detected sensor value.
The sensor monitoring method according to claim 6, further comprising: selecting a sensor with less disorder in the correlation from the plurality of sensor sets.
The sensor monitoring method according to claim 6 or 7, wherein the plurality of sensor sets are attached to different physical positions to obtain sensor detection values.
The sensor monitoring method according to claim 6 or 7, wherein the plurality of sensor sets are attached to different physical positions of the power plant to obtain sensor detection values.
10. The sensor monitoring method according to claim 6, further comprising: estimating a drift amount in each sensor of the plurality of sensor sets, and performing correction based on the estimated drift amount.
A sensor monitoring program executed at a power plant,
A process of referring to a database storing a relationship model in which a correlation between sensor values from a plurality of sensors constituting each sensor set and sensor values from a plurality of sensors constituting another sensor set is modeled;
Correlation between sensor detection values in the plurality of sensors constituting each sensor set and sensor detection values in the plurality of sensors constituting another sensor set, and the modeled relationship stored in the database A process of comparing the correlation of the models;
A sensor monitoring program for causing a computer to execute a process of detecting a disturbance from the correlation of the modeled relation model in the correlation of the detected sensor values.