Plant abnormality diagnosis system

Abstract

A plant abnormality diagnosis system comprising: a first score calculation unit that is configured to calculate, for each sensor among a plurality of types of sensors that are provided in a plant, a first score that indicates the degree of abnormality in actual measurement values from the sensor; an overall evaluation score calculation unit that is configured to calculate, for each abnormal event among a plurality of types of abnormal events that can occur at the plant, an overall evaluation score that indicates the magnitude of the possibility that the abnormal event is occurring on the basis of the first score calculated for each sensor by the first score calculation unit and weighting coefficients that indicate, for each sensor, the degree of relatedness between the sensor and each of the abnormal events; and a diagnosis results output unit that is configured to extract candidates for abnormal events occurring at the plant from the plurality of types of abnormal events on the basis of the overall evaluation score calculated for each abnormal event by the overall evaluation score calculation unit and to output information including the extracted candidates for abnormal events as diagnosis results for the plant.

Description

Plant abnormality diagnosis system

[0001] The present disclosure relates to a plant abnormality diagnosis system.

[0002] Patent Literature 1 discloses a monitoring and diagnostic system that monitors a plant and diagnoses abnormalities. This monitoring and diagnostic system refers to a judgment and evaluation result database to sequentially read differences between thresholds in plant operation data, and then refers to a database of abnormality signs and abnormal event related information. At this time, information such as the identification of the abnormal event, the extent of the abnormality, and recommended measures is obtained as events that cause the difference from the thresholds using parameters linked to the operation data, and this information is registered as diagnostic result data in a diagnostic result database.

[0003] Japanese Patent Application Laid-Open No. 2020-107025

[0004] Incidentally, multiple types of abnormal events can occur in a plant, and it is necessary to accurately diagnose the abnormal event that is actually occurring from among those abnormal events. However, the system described in Patent Document 1 has room for improvement in terms of the accuracy of diagnosing abnormal events.

[0005] In view of the above circumstances, at least one embodiment of the present disclosure has an object to provide a plant abnormality diagnosis system that can accurately diagnose abnormal events in a plant.

[0006] In order to achieve the above object, a plant abnormality diagnosis system according to at least one embodiment of the present disclosure is a plant abnormality diagnosis system for diagnosing an abnormal event occurring in a plant, comprising: a first score calculation unit configured to calculate, for each of a plurality of types of sensors provided in the plant, a first score indicating a degree of abnormality in an actual measurement value of the sensor; a comprehensive evaluation score calculation unit configured to calculate, for each of a plurality of types of abnormal events that may occur in the plant, an overall evaluation score indicating a degree of possibility that the abnormal event has occurred, based on a weight coefficient indicating, for each of the sensors, a degree of association between each of the abnormal events and the sensor and the first score calculated by the first score calculation unit; and a diagnosis result output unit configured to extract candidates for the abnormal event occurring in the plant from the plurality of types of abnormal events, based on the overall evaluation score for each of the abnormal events calculated by the comprehensive evaluation score calculation unit, and output information including the extracted candidates for the abnormal event as a diagnosis result for the plant.

[0007] According to at least one embodiment of the present disclosure, a plant abnormality diagnosis system capable of accurately diagnosing abnormal events in a plant is provided.

[0008] 5A is a diagram showing a schematic configuration of a thermal power plant 4 that is a target of diagnosis by a plant abnormality diagnosis system 2 according to an embodiment. FIG. 5B is a diagram showing an example of the hardware configuration of the plant abnormality diagnosis system 2. FIG. 5C is a block diagram showing an example of the functional configuration of the plant abnormality diagnosis system 2 shown in FIGS. 1 and 2. FIG. 5D is a diagram showing a flow of abnormality diagnosis by the plant abnormality diagnosis system 2 shown in FIG. 3. FIG. 5E is a diagram for explaining an example of a sign of abnormality in S103 of FIG. 3, showing a comparison between an actual measurement value and a predicted value of a sensor 6 that measures a field current of a generator G of the plant 4. FIG. 5F is a graph showing the difference between the actual measurement value and the predicted value of the sensor 6 shown in FIG. 5A, and thresholds (upper and lower limit values) used in S103. FIG. 5G is a diagram showing an example of a weighting coefficient table E. FIG. 5H is a diagram for explaining a method of calculating a comprehensive evaluation score S3 using a first score S1 and a weighting coefficient W. FIG. 5I is an example of information indicating a diagnosis result of the plant 4 that is output from the diagnosis result output unit 18. FIG. 5I is a table showing, for each abnormal event, a specific sensor 6 that specifies a necessary condition for the diagnosis result output unit 18 to extract candidates for an abnormal event as information indicating the diagnosis result.

[0009] Several embodiments of the present disclosure will be described below with reference to the accompanying drawings. However, the dimensions, materials, shapes, relative arrangements, etc. of components described as embodiments or shown in the drawings are merely illustrative examples and are not intended to limit the scope of the invention. For example, expressions expressing relative or absolute arrangements, such as "in a certain direction," "along a certain direction," "parallel," "orthogonal," "center," "concentric," or "coaxial," not only strictly express such arrangements, but also express relative displacements with a tolerance or angle or distance to the extent that the same function is achieved. For example, expressions expressing the equality of things, such as "same," "equal," and "homogeneous," not only express strict equality, but also express tolerance or differences to the extent that the same function is achieved. For example, expressions expressing shapes such as a square or cylindrical shape not only express shapes such as a square or cylindrical shape in the strict geometric sense, but also express shapes including concave and convex portions, chamfered portions, etc., to the extent that the same effect is achieved. On the other hand, the expressions "comprise," "include," "have," "includes," or "have" of one element are not exclusive expressions that exclude the presence of other elements.

[0010] Fig. 1 is a diagram showing a schematic configuration of a thermal power plant 4 that is a target of diagnosis by a plant anomaly diagnosis system 2 according to one embodiment. The thermal power plant 4 shown in Fig. 1 (hereinafter simply referred to as "plant 4") includes, as main equipment, a boiler B, a turbine T, and a generator G, and further includes, as large-scale auxiliary equipment, a boiler auxiliary B1, a turbine auxiliary T1, a generator auxiliary G1, a condenser C, and a generator excitation circuit G2. Although not shown in Fig. 1, various sensors installed in the thermal power plant 4 detect physical quantities such as various process quantities and state quantities, and operation data D1 including the detected physical quantities is input into the plant anomaly diagnosis system 2.

[0011] 2 is a diagram illustrating an example of the hardware configuration of the plant abnormality diagnosis system 2. As illustrated in FIG. 2 , the plant abnormality diagnosis system 2 includes, for example, a processor 72, a random access memory (RAM) 74, a read-only memory (ROM) 76, a hard disk drive (HDD) 78, an input I / F 80, and an output I / F 82, which are connected to one another via a bus 84, and is configured using a computer. Note that the hardware configuration of the plant abnormality diagnosis system 2 is not limited to the above, and may be configured using a combination of a control circuit and a storage device. The plant abnormality diagnosis system 2 is also configured by a computer executing a program that realizes each function of the plant abnormality diagnosis system 2. The functions of each part of the plant abnormality diagnosis system 2 described below are realized, for example, by loading a program stored in the ROM 76 into the RAM 74 and executing it with the processor 72, and by reading and writing data from and to the RAM 74 and the ROM 76. The hardware constituting the plant abnormality diagnosis system 2 may be concentrated in one location, or may be distributed across multiple locations.

[0012] Fig. 3 is a block diagram showing an example of the functional configuration of the plant abnormality diagnosis system 2 shown in Fig. 1 and Fig. 2. Fig. 4 is a diagram showing a flow of abnormality diagnosis by the plant abnormality diagnosis system 2 shown in Fig. 3.

[0013] As shown in FIG. 3 , the plant abnormality diagnosis system 2 includes a data acquisition unit 10, an abnormality sign detection unit 12, a first score calculation unit 14, a comprehensive evaluation score calculation unit 16, a diagnosis result output unit 18, a memory unit 20, and a display unit 22.

[0014] 4 , in S101, the data acquiring unit 10 acquires operation data D1 of the plant 4 from multiple types of sensors 6 installed in the plant 4. The operation data D1 here includes time-series data of actual values ​​of multiple types of physical quantities measured by the multiple types of sensors 6 in multiple devices (e.g., the above-mentioned generator G and generator auxiliary G1) installed in the plant 4. Furthermore, examples of the multiple types of physical quantities measured by the multiple types of sensors 6 for the generator G of the plant 4 include active power, reactive power, field current, field voltage, field coil resistance, power factor, and armature current.

[0015] Next, in S102, the abnormality precursor detection unit 12 calculates, for each of multiple types of sensors 6, the difference between the actual measured value of the physical quantity measured by the sensor 6 and the predicted value of the output of the sensor 6 (the physical quantity measured by the sensor 6) predicted using multiple regression analysis.

[0016] In this multiple regression analysis, a prediction model M for predicting a response variable (dependent variable) from a plurality of explanatory variables (independent variables) of the plant 4 is created in advance based on operation data D1 of the plant 4 when the plant 4 is operating normally (operation data existing in a normal space), and the prediction model M is stored in the storage unit 20. The abnormality sign detection unit 12 then inputs the operation data D1 to be subjected to abnormality diagnosis into the prediction model M read from the storage unit 20, and calculates predicted values ​​of the outputs of the plurality of sensors 6 using the prediction model M. As described above, the abnormality sign detection unit 12 then calculates, for each sensor 6, the difference (i.e., the value obtained by subtracting the predicted value from the actual measurement value of the sensor 6) between the actual measurement value of the sensor 6 and the predicted value of the output of the sensor 6 predicted using the multiple regression analysis (i.e., the predicted value of the output of the sensor 6 calculated using the prediction model M).

[0017] In this specification, including the following explanation, the term "actual measured value" means the value of a physical quantity measured by the sensor 6, and the term "predicted value" means the predicted value of the output of the sensor 6 calculated based on the operating data D1 in S102.

[0018] Next, in S103, if the difference calculated in S102 for at least one type of sensor 6 out of the multiple types of sensors 6 (i.e., the difference between the actual measurement value and the predicted value of the sensor 6) exceeds a threshold, the abnormality sign detection unit 12 determines that there is a sign of abnormality in the plant 4 (detects a sign of abnormality), and proceeds to S104. In S103, if the difference calculated in S102 for none of the multiple types of sensors 6 exceeds the threshold, the abnormality sign detection unit 12 determines that there is no sign of abnormality in the plant 4, and returns to S101. Note that the threshold here may be an upper limit value (positive threshold value) and a lower limit value (negative threshold value) for defining a normal range of the value of the sensor 6, and is determined for each sensor 6 and stored in the storage unit 20. In this case, the abnormality precursor detection unit 12 may determine that there is a sign of an abnormality in the plant 4 when the difference calculated in S102 for at least one type of sensor 6 out of the multiple types of sensors 6 exceeds the upper limit value (exceeds the upper limit value in a positive direction) or falls below the lower limit value (exceeds the lower limit value in a negative direction).

[0019] 5A is a diagram for explaining an example of a sign of an abnormality in S103, and is a diagram showing time-series data of actual measured values ​​and predicted values ​​of a sensor 6 that measures the field current of the generator G of the plant 4. FIG. 5B is a graph showing time-series data of the difference between the actual measured value and the predicted value of the sensor 6 shown in FIG. 5A and threshold values ​​(upper limit and lower limit) used in S103. In the examples shown in FIGS. 5A and 5B, immediately after the shutdown period of the generator G ends and the generator G is started up, the actual measured value of the sensor 6 significantly deviates from the predicted value, and the difference between the actual measured value and the predicted value exceeds the threshold value (the upper limit in the example shown in FIG. 5B). Therefore, the abnormality sign detection unit 12 determines that there is a sign of an abnormality in the plant 4.

[0020] 4 , in S104, the first score calculation unit 14 calculates, for each of the multiple types of sensors 6 provided in the plant 4, a first score S1 indicating the degree of abnormality of the actual measurement value of the sensor 6. For example, the first score calculation unit 14 may calculate, as the first score S1, a value that has a positive correlation with the difference between the actual measurement value of the sensor 6 calculated for each sensor 6 in S102 and the predicted value (for example, a value proportional to the difference), or may calculate, as the first score S1, the absolute value of the value obtained by dividing the difference between the actual measurement value of the sensor 6 calculated for each sensor 6 in S102 and the predicted value by the threshold value set for each sensor (the upper limit value or the lower limit value used in S103).

[0021] In S105, the comprehensive evaluation score calculation unit 16 calculates a sensor-specific evaluation score S2 for each of multiple types of abnormal events that may occur in the plant 4, which is a value (i.e., W x S1) obtained by multiplying, for each sensor 6, a weighting coefficient W that indicates the degree of association between the abnormal event and the sensor 6 and the first score S1 for each sensor 6 calculated by the first score calculation unit 14. Note that, in the case of the generator G of the plant 4, the multiple types of abnormal events here include, for example, an abnormality in an instrument or transmitter, a sudden change in the state of the power grid, a rare short in the field winding, a change in rotor balance, excessive turbine output, and an abnormality in the lubricating oil pressure control valve. In the example shown in FIG. 6 , the weighting coefficient W ranges from 0 to 5, and the larger the value, the stronger the degree of association between the abnormal event and the sensor. In addition, the memory unit 20 may store a weighting coefficient table E (see Figure 6) that specifies a weighting coefficient W that indicates the degree of association between each abnormal event and the sensor 6 for each sensor 6. In this case, the overall evaluation score calculation unit 16 refers to the weighting coefficient table E stored in the memory unit 20 and multiplies the weighting coefficient W that indicates the degree of association between each abnormal event and the sensor 6 for each sensor 6 by the first score S1 for each sensor 6 calculated by the first score calculation unit 14 for each sensor 6, thereby calculating a sensor-specific evaluation score S2 for each abnormal event for each sensor 6, as shown in Figure 7.

[0022] 6 and 7 , the weighting coefficients are set to be different as necessary for a case where the difference between the actual measurement value of the sensor 6 and the predicted value exceeds an upper limit (when the output of the sensor 6 is excessively large compared to the predicted value) and a case where the difference between the actual measurement value of the sensor 6 and the predicted value is below a lower limit (when the output of the sensor 6 is too small compared to the predicted value). For example, when abnormal event B is "excessive turbine output," the actual measurement values ​​of the active power, reactive power, etc. of the generator G will not be too small compared to the predicted values. Therefore, the weighting coefficient W indicating the degree of association between the sensor 6 (e.g., sensors 6a to 6c in the example shown in FIG. 7 ) that measures the active power, reactive power, etc. of the generator G and abnormal event B may be set to 1 only when the difference between the actual measurement value of the sensor 6 and the predicted value exceeds the upper limit, and may be set to 0 when the difference between the actual measurement value of the sensor 6 and the predicted value is below the lower limit.

[0023] Next, in S106, the overall evaluation score calculation unit 16 calculates an overall evaluation score S3 for each abnormal event by summing, for each abnormal event, the sensor-specific evaluation scores S2 of the multiple types of sensors 6 calculated in S105. For example, in the example shown in Fig. 7 , for abnormal event A, the overall evaluation score calculation unit 16 calculates an overall evaluation score S3 of 48.8 for the abnormal event A by summing the sensor-specific evaluation score S2 of sensor 6a (0), the sensor-specific evaluation score S2 of sensor 6b (15.5), the sensor-specific evaluation score S2 of sensor 6c (0), the sensor-specific evaluation score S2 of sensor 6d (13.0), and the sensor-specific evaluation scores S2 of all the other sensors 6. Similarly, for each of the other abnormal events, the overall evaluation score S3 for each abnormal event is calculated by summing the sensor-specific evaluation scores S2 of the multiple types of sensors 6 for each abnormal event.

[0024] Next, in S107, the diagnosis result output unit 18 extracts candidates for abnormal events occurring in the plant 4 from among the multiple types of abnormal events based on the overall evaluation score S3 for each abnormal event calculated by the overall evaluation score calculation unit 16, and outputs information including the extracted candidates for abnormal events as a diagnosis result for the plant 4. For example, the diagnosis result output unit 18 may extract one or more abnormal events from among the multiple types of abnormal events in descending order of overall evaluation score S3 based on the overall evaluation score S3 for each abnormal event calculated by the overall evaluation score calculation unit 16, or may extract multiple abnormal events from among the multiple types of abnormal events in descending order of overall evaluation score S3.

[0025] FIG. 8 illustrates an example of information indicating the diagnosis result of the plant 4 output in S107. The diagnosis result illustrated in FIG. 8 includes the top three abnormal events ranked in descending order of overall evaluation score S3 from among multiple types of abnormal events, and recommended countermeasures for each of the abnormal events. In the illustrated example, the diagnosis result indicates the first, second, and third most likely candidates for the abnormal event, respectively, as "instrument or transmitter abnormality," "field winding rare short," and "change in rotor balance." The countermeasures for each abnormal event are also indicated as "inspection of instruments and transmitters," "inspection for rare shorts," and "inspection for changes in shaft vibration." The diagnosis result output unit 18 may be configured to output information indicating the diagnosis result to a display unit 22 (see FIG. 3 ), such as a display, to display the diagnosis result on the display unit 22.

[0026] According to the plant abnormality diagnosis system 2, an overall evaluation score S3 is calculated based on a weighting coefficient W indicating, for each sensor 6, the degree of association between each abnormal event and the sensor 6, and a first score S1 indicated for each sensor 6 that indicates the degree of abnormality in the actual measurement value of the sensor 6, and candidates for abnormality diagnosis are extracted based on the overall evaluation score S3, thereby making it possible to accurately diagnose abnormal events in the plant 4. The inventors of the present application have confirmed that the diagnosis results of the plant abnormality diagnosis system 2 are generally similar to the analysis results and diagnosis examples by experts, and even those who do not have advanced skills in abnormality diagnosis of the plant 4 can accurately diagnose abnormal events in the plant 4 by using the plant abnormality diagnosis system 2. Furthermore, according to the plant abnormality diagnosis system 2, it is possible to accurately diagnose abnormal events in each piece of equipment, such as the generator G and generator auxiliary machine G1, of the plant 4.

[0027] In some embodiments, the diagnostic result output unit 18 may be configured to exclude a specific abnormal event associated with a specific sensor 6 from among the multiple types of abnormal events from the candidate abnormal events in the diagnostic result, regardless of the overall evaluation score S3 of the specific abnormal event, when the difference between the actual measurement value and the predicted value of the specific sensor 6 from among the multiple types of sensors 6 does not exceed a threshold. In this case, as shown in Fig. 9, the storage unit 20 (see Fig. 3) may store information indicating necessary conditions for each of the multiple types of abnormal events to be extracted as the candidate abnormal event by the diagnostic result output unit 18.

[0028] 9 , the necessary conditions stored in the storage unit 20 include the difference between the actual measurement value and the predicted value of a specific sensor 6 (sensor 6b in the illustrated example) among the multiple types of sensors 6 exceeding a threshold. In the example illustrated in FIG. 9 , for each abnormal event, a number 1 or 0 is assigned to each sensor 6 as information for identifying whether the sensor 6 is the specific sensor. The specific sensor 6 is assigned the number 1 indicating that it is the specific sensor 6, and the sensor 6 that is not the specific sensor 6 is assigned the number 0 indicating that it is not the specific sensor 6. In the example illustrated in FIG. 9 , when abnormal event B, which is the specific abnormal event, occurs in the plant 4, it necessarily means that the difference between the actual measurement value and the predicted value of sensor 6b exceeds a threshold (upper limit value), and a necessary condition for abnormal event B to be extracted as a candidate by the diagnosis result output unit 18 includes the difference between the actual measurement value and the predicted value of sensor 6b, which is the specific sensor 6, exceeding a threshold (exceeding the upper limit value).

[0029] For example, if abnormal event B is a rare short circuit in the field winding and sensor 6b is a sensor that detects the field current, when a rare short circuit in the field winding occurs, the difference between the actual measurement value and the predicted value of sensor 6b that detects the field current must necessarily exceed the threshold. Therefore, if the difference between the actual measurement value and the predicted value of sensor 6b does not exceed the threshold, the diagnostic result output unit 18 does not extract abnormal event B as a candidate for the abnormal event, regardless of the overall evaluation score S3 of abnormal event B. Only when the difference between the actual measurement value and the predicted value of sensor 6b exceeds the threshold, and if the ranking of abnormal event B, determined according to the overall evaluation score S3, is within a predetermined ranking (e.g., within the top three), does the diagnostic result output unit 18 extract abnormal event B as a candidate for the abnormal event.

[0030] In the example shown in Figure 9, the information (1 or 0) for identifying whether the sensor 6 is the specific sensor 6 is changed as necessary depending on whether the difference between the actual measurement value and the predicted value of the sensor 6 exceeds the upper limit value (when the output of the sensor 6 is excessive) or whether the difference between the actual measurement value and the predicted value of the sensor 6 is below the lower limit value (when the output of the sensor 6 is too small).

[0031] 9 , when the difference between the actual measurement value and the predicted value of the specific sensor 6 b does not exceed the threshold, the diagnostic result output unit 18 excludes the abnormal event B from the candidates for the abnormal event in the diagnostic result, regardless of the overall evaluation score S3 of the abnormal event B associated with the specific sensor 6 b. Therefore, when the diagnostic result output unit 18 extracts candidates for the abnormal event, it can appropriately narrow down the candidates for the abnormal event, and it is possible to more accurately diagnose abnormal events in the plant 4.

[0032] The present disclosure is not limited to the above-described embodiments, but also includes modifications to the above-described embodiments and appropriate combinations of these modifications.

[0033] For example, in the above embodiment, the abnormality precursor detection unit 12 used multiple regression analysis to calculate the predicted value of the sensor 6, but the method for calculating the predicted value of the sensor 6 is not limited to multiple regression analysis and may also be, for example, the Mahalanobis-Taguchi method (MT method).

[0034] In the example shown in FIG. 4, steps S104 to S107 are executed when the difference between the actual measurement value and the predicted value of the sensor 6 exceeds a threshold value. However, the trigger for executing steps S104 to S107 is not limited to the difference between the actual measurement value and the predicted value of the sensor 6 exceeding a threshold value, and steps S104 to S107 may be executed periodically at predetermined intervals, for example.

[0035] In addition, in some embodiments, the abnormality precursor detection unit 12 may be configured to be able to change the threshold value used in S103, and when this threshold value is used to calculate the first score S1, the threshold value used to calculate the first score S1 may also be able to be changed.

[0036] The contents described in each of the above embodiments can be understood, for example, as follows.

[0037] [1] A plant abnormality diagnosis system according to at least one embodiment of the present disclosure (e.g., the above-described plant abnormality diagnosis system 2) is a plant abnormality diagnosis system for diagnosing an abnormal event occurring in a plant (e.g., the above-described plant 4), and includes: a first score calculation unit (e.g., the above-described first score calculation unit 14) configured to calculate, for each of a plurality of types of sensors (e.g., the above-described plurality of types of sensors 6) provided in the plant, a first score (e.g., the above-described first score S1) indicating the degree of abnormality of an actual measurement value of the sensor; a comprehensive evaluation score calculation unit (e.g., the above-described comprehensive evaluation score calculation unit 16) configured to calculate, for each of a plurality of types of abnormal events that may occur in the plant, an overall evaluation score (e.g., the above-described comprehensive evaluation score S3) indicating the magnitude of the possibility that the abnormal event has occurred, based on a weight coefficient (e.g., the above-described weight coefficient W) indicating, for each of the sensors, the degree of association between each of the abnormal events and the sensor, and the first score calculated by the first score calculation unit for each of the sensors; and a diagnostic result output unit (for example, the above-mentioned diagnostic result output unit 18) configured to extract candidates for the abnormal event occurring in the plant from the plurality of types of abnormal events based on the overall evaluation score for each of the abnormal events calculated by the overall evaluation score calculation unit, and to output information including the extracted candidates for the abnormal event as a diagnostic result for the plant.

[0038] According to the plant abnormality diagnosis system described in [1] above, an overall evaluation score is calculated based on a weighting coefficient indicating, for each sensor, the degree of association between each abnormal event and the sensor, and a first score indicated for each sensor indicating the degree of abnormality of the actual measured value of the sensor, and candidates for abnormality diagnosis are extracted based on the overall evaluation score, thereby making it possible to accurately diagnose abnormal events in the plant.

[0039] [2] In some embodiments, in the plant abnormality diagnosis system described in [1] above, the diagnosis result output unit is configured to extract, as one of the candidates, the abnormal event having the largest overall evaluation score from among the multiple types of abnormal events, based on the overall evaluation score for each of the abnormal events calculated by the overall evaluation score calculation unit.

[0040] According to the plant abnormality diagnosis system described in [2] above, the abnormal event with the highest overall evaluation score among multiple types of abnormal events is extracted as one of the candidates, thereby making it possible to accurately diagnose abnormal events in the plant.

[0041] [3] In some embodiments, in the plant abnormality diagnosis system described in [1] or [2] above, the overall evaluation score calculation unit is configured to calculate, for each of the plurality of types of abnormal events, a sensor-specific evaluation score (e.g., the above-mentioned sensor-specific evaluation score S2) which is a value obtained by multiplying, for each of the plurality of types of abnormal events, the weight coefficient indicating the degree of association between each of the abnormal events and the sensor for each of the sensors by the first score for each of the sensors calculated by the first score calculation unit, and further to calculate the overall evaluation score for each of the abnormal events based on the sensor-specific evaluation scores of the plurality of types of sensors.

[0042] According to the plant abnormality diagnosis system described in [3] above, an overall evaluation score is calculated for each abnormal event based on the sensor-specific evaluation score obtained by multiplying the weighting coefficient and the first score for each sensor, thereby making it possible to accurately diagnose abnormal events in the plant.

[0043] [4] In some embodiments, in the plant abnormality diagnosis system described in [3] above, the overall evaluation score calculation unit is configured to calculate the overall evaluation score for each abnormal event by summing the sensor-specific evaluation scores of the multiple types of sensors for each abnormal event.

[0044] According to the plant abnormality diagnosis system described in [4] above, abnormal events in the plant can be accurately diagnosed by adding up the sensor-specific evaluation scores of multiple types of sensors for each abnormal event to calculate an overall evaluation score for each abnormal event.

[0045] [5] In some embodiments, in the plant abnormality diagnosis system described in any of [1] to [4] above, the first score calculation unit is configured to calculate the first score for each sensor based on the difference between the actual measured value and the predicted value of the sensor.

[0046] According to the plant abnormality diagnosis system described in [5] above, a first score indicating the degree of abnormality of the actual measurement value of the sensor can be calculated based on the difference between the actual measurement value of the sensor and the predicted value.

[0047] [6] In some embodiments, in the plant abnormality diagnosis system described in [5] above, the diagnosis result output unit is configured to exclude a specific abnormal event (e.g., abnormal event B in the example shown in FIG. 9 ) associated with a specific sensor among the multiple types of abnormal events from the candidates for the abnormal event in the diagnosis result, regardless of the overall evaluation score of the specific abnormal event, when the difference between the actual measurement value and the predicted value of a specific sensor (e.g., sensor 6 b in the example shown in FIG. 9 ) among the multiple types of sensors does not exceed a threshold value.

[0048] Among the multiple types of abnormal events, there is an abnormal event (the specific abnormal event) in which, when a certain abnormal event (the specific abnormal event) occurs, the difference between the actual measurement value and the predicted value of a specific sensor always exceeds a threshold. For example, when an abnormal event such as a layer short in a field winding occurs in a generator of a plant, the difference between the actual measurement value and the predicted value of a sensor that detects field current always exceeds a threshold.

[0049] For this reason, in the plant abnormality diagnosis system described in [6] above, for such a specific abnormal event, the difference between the actual measurement value and the predicted value of a specific sensor out of multiple types of sensors exceeds a threshold, which is a necessary condition for the specific abnormal event to be extracted as the candidate by the diagnosis result output unit.If the difference between the actual measurement value and the predicted value of the specific sensor does not exceed the threshold, the diagnosis result output unit will not extract the specific abnormal event as the candidate, regardless of the overall evaluation score of the specific abnormal event.Therefore, when the diagnosis result output unit extracts candidates for abnormal events, it can appropriately narrow down the candidates for abnormal events, and it is possible to more accurately diagnose abnormal events in the plant.

[0050] DESCRIPTION OF SYMBOLS 2 Plant abnormality diagnosis system 4 Plant 6, 6a, 6b, 6c, 6d Sensor 10 Data acquisition unit 12 Abnormality sign detection unit 14 First score calculation unit 16 Overall evaluation score calculation unit 18 Diagnosis result output unit 20 Memory unit 22 Display unit 72 Processor 74 RAM 76 ROM 78 HDD 80 Input I / F 82 Output I / F 84 Bus B Boiler B1 Boiler auxiliary equipment C Condenser D1 Operation data E Weighting coefficient table G Generator G1 Generator auxiliary equipment G2 Generator excitation circuit M Prediction model S1 First score S2 Evaluation score by sensor S3 Overall evaluation score T Turbine T1 Turbine auxiliary equipment W Weighting coefficient

Claims

1. A plant abnormality diagnosis system for diagnosing abnormal events occurring in a plant, comprising: a first score calculation unit configured to calculate, for each of a plurality of types of sensors provided in the plant, a first score indicating the degree of abnormality of the measured value of the sensor; a comprehensive evaluation score calculation unit configured to calculate, for each of a plurality of types of abnormal events that may occur in the plant, a comprehensive evaluation score indicating the likelihood of occurrence of the abnormal event based on a weight coefficient indicating the degree of association between each of the abnormal events and the sensor for each of the sensors and the first score for each of the sensors calculated by the first score calculation unit; and a diagnosis result output unit configured to extract, based on the comprehensive evaluation score for each of the abnormal events calculated by the comprehensive evaluation score calculation unit, candidates for the abnormal events occurring in the plant from among the plurality of types of abnormal events, and output information including the extracted candidates for the abnormal events as a diagnosis result of the plant.

2. The plant abnormality diagnosis system according to claim 1, wherein the diagnosis result output unit is configured to extract, as one of the candidates, the abnormal event having the highest comprehensive evaluation score among the plurality of types of abnormal events based on the comprehensive evaluation score for each of the abnormal events calculated by the comprehensive evaluation score calculation unit.

3. The plant abnormality diagnosis system according to claim 1, wherein the comprehensive evaluation score calculation unit calculates, for each of the plurality of types of abnormal events, a sensor-specific evaluation score that is a value obtained by multiplying, for each of the sensors, the weight coefficient indicating the degree of association between each of the abnormal events and the sensor and the first score for each of the sensors calculated by the first score calculation unit, and further calculates the comprehensive evaluation score for each of the abnormal events based on the sensor-specific evaluation scores of the plurality of types of sensors.

4. The plant abnormality diagnosis system according to claim 3, wherein the comprehensive evaluation score calculation unit is configured to calculate the comprehensive evaluation score for each of the abnormal events by summing the sensor-specific evaluation scores of the plurality of types of sensors for each of the abnormal events.

5. The plant abnormality diagnosis system according to claim 1, wherein the first score calculation unit is configured to calculate the first score for each of the sensors based on a difference between the measured value and the predicted value of the sensor.

6. The plant abnormality diagnosis system according to claim 5, wherein when a difference between the measured value and the predicted value of a specific sensor among the plurality of types of sensors does not exceed a threshold value, the diagnosis result output unit excludes a specific abnormality event associated with the specific sensor from candidates of the abnormality event in the diagnosis result regardless of the comprehensive evaluation score of the specific abnormality event.

Images