State monitoring method for plant appliance and state monitoring system for plant appliance

Abstract

To provide a state monitoring method for a plant appliance and a state monitoring system for the plant appliance that can diagnose an abnormality in the plant appliance with high accuracy.SOLUTION: A state monitoring method for a plant appliance of the present invention for monitoring a state of the plant appliance, comprises: a reference data generating step of generating reference data by excluding data from measurement data when the one parameter value exceeds a second threshold value in the case where a data changing amount exceeds a first threshold value in a second time zone included in a first time zone of measurement data of a state of the plant appliance, and one parameter value exceeds the second threshold value in a second time zone; and a step of comparing the monitoring measurement data for the state of the plant appliance with the reference data to monitor the state of the appliance.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a plant equipment status monitoring method and a plant equipment status monitoring system. [Background technology]

[0002] Chemical plants and power plants are increasingly introducing state monitoring and diagnostic technologies for plant equipment with the aim of stable operation and streamlining maintenance work. Early detection and countermeasures for abnormalities in plant equipment can be expected to enable continued operation and shorten maintenance work time. One method for monitoring the state of plant equipment uses plant measurement values ​​such as temperature, pressure, vibration, and rotation speed. Measurement data from the period when the equipment is functioning normally is used as reference data, and measurement data from the period being diagnosed is used as diagnostic data, and the presence or absence of an abnormality is determined by comparing the reference data with the diagnostic data.

[0003] As an example of such a condition monitoring and diagnosis technology, Patent Document 1 discloses the following technology: "The parameter abnormality detection unit determines whether or not the detection values ​​(time series data stored in the time series data storage unit) of the pressure sensor, temperature sensor, and flow rate sensor acquired by the detection value acquisition device contain abnormal detection values, based on the normal data described above."

[0004] Patent Document 2 discloses a technology for "calculating the degree of abnormality of a group of monitoring-time feature quantities, which are a plurality of feature quantities extracted from the state quantity fluctuation data of the monitored equipment when the monitored equipment is being monitored, based on a group of normal-time feature quantities, which are a plurality of feature quantities extracted from the state quantity fluctuation data of the monitored equipment when the monitored equipment is in a normal state." [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2021-174195 [Patent Document 2] Japanese Patent Application Publication No. 2019-128704 Summary of the Invention [Problem to be solved by the invention]

[0006] For large rotating machinery such as compressors used in various plants, the equipment configuration, connections, and measurement parameters vary from plant to plant, and there are only a limited number of devices that can be compared under the same conditions. Furthermore, in actual plants, measurement data that can be considered completely normal is often limited and unavailable. Even if measurement data that can be considered normal is available, it may only be obtained over a short period of time. In order to accurately diagnose abnormalities using limited measurement data as reference data, it is important to effectively utilize measurement data that can serve as reference data.

[0007] However, if you try to diagnose the state during steady operation as diagnostic data while including instantaneous fluctuations in temperature or transient states of a process such as start-up and shutdown in the reference data range, the instantaneous fluctuations and transient states of the reference data may be considered normal, which may reduce the accuracy of abnormality detection.

[0008] An object of the present invention is to provide a plant equipment status monitoring method and a plant equipment status monitoring system that can diagnose abnormalities in plant equipment with high accuracy. [Means for solving the problem]

[0009] The plant equipment status monitoring method of the present invention is characterized in that, in the plant equipment status monitoring method, when a data change amount exceeds a first threshold value during a second time period included in a first time period of measurement data of the plant equipment status and one parameter value exceeds a second threshold value during the second time period, the method comprises a reference data generation step of generating reference data by excluding from the measurement data the data for a time period during which the one parameter value exceeded the second threshold value, and a step of comparing the measurement data for monitoring the plant equipment status with the reference data to monitor the equipment status.

[0010] Alternatively, the plant equipment status monitoring system of the present invention is characterized in that, in the plant equipment status monitoring system that monitors the status of the plant equipment, it comprises a reference data generation unit that, when a data change amount exceeds a first threshold value in a second time period included in a first time period of measurement data of the status of the plant equipment and one parameter value exceeds a second threshold value in the second time period, creates reference data by excluding from the measurement data the data for the time period in which the one parameter value exceeded the second threshold value, and an abnormality degree calculation unit that compares the measurement data for monitoring the status of the plant equipment with the reference data to calculate the degree of abnormality. [Effects of the Invention]

[0011] According to the present invention, it is possible to provide a plant equipment state monitoring method and a plant equipment state monitoring system that can diagnose abnormalities in plant equipment with high accuracy. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a configuration diagram of a condition monitoring and diagnosis system according to an embodiment of the present invention; [Figure 2] 1 shows a centrifugal compressor train configuration as an application example of the condition monitoring and diagnosis of this embodiment. [Figure 3] 10 is a diagram illustrating an example of measured values ​​in the state monitoring diagnosis of the present embodiment. [Figure 4] 10 shows an example of one parameter of the first reference data for the condition monitoring and diagnosis of this embodiment. [Figure 5] FIG. 2 is a diagram illustrating a configuration of a reference data generating unit for the condition monitoring and diagnosis according to the present embodiment. [Figure 6] 10 is a diagram showing an example of first reference data and second time periods of the condition monitoring and diagnosis of the present embodiment, an abnormality level for each second time period, and one parameter value. FIG. [Figure 7] 10 shows an example of reference data for the condition monitoring and diagnosis of this embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0013] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. The same reference numerals are used throughout the drawings to designate the same components. [Example]

[0014] Fig. 1 is a configuration diagram of a condition monitoring and diagnostic system of this embodiment. In the condition monitoring and diagnostic system 1, a measurement value input unit 2 inputs measurement values ​​from sensors installed in equipment. In this embodiment, a centrifugal compressor, which is a representative plant equipment, is the subject of the test. Here, the subject centrifugal compressor train 200 has a configuration as shown in Fig. 2, in which a centrifugal compressor 201 and a driver 202 are connected by a coupling 203.

[0015] The centrifugal compressor 201 is provided with sensors such as four axial vibration sensors SC1 to SC4 that monitor the axial vibration of the centrifugal compressor rotor 204, bearing temperature sensors SC5 and SC6 for the bearings that rotatably support the centrifugal compressor rotor 204, SC7 that monitors the rotation speed, a pressure sensor SC8 that monitors the suction pressure, a pressure sensor SC9 that monitors the discharge pressure, and a flow rate sensor SC10 that monitors the flow rate. The axial vibration sensors SC1 and SC2 are arranged, for example, at 90-degree intervals in the circumferential direction of the centrifugal compressor 201. The axial vibration sensors SC3 and SC4 are arranged, for example, at 90-degree intervals in the circumferential direction of the centrifugal compressor 201.

[0016] The driver 202 is provided with sensors such as four shaft vibration sensors SD1 to SD4 that monitor the shaft vibration of the driver rotor 205. Although only some of the sensors are described here, various other sensors are usually provided in addition to these. The shaft vibration sensors SD1 and SD2 are arranged, for example, at 90-degree intervals in the circumferential direction of the driver 202. The shaft vibration sensors SD3 and SD4 are arranged, for example, at 90-degree intervals in the circumferential direction of the driver 202.

[0017] In Fig. 1, a measurement value input unit 2 inputs measurement values ​​such as shaft vibration, suction pressure, discharge pressure, and rotation speed obtained from sensors SC1 to SC9, SD1 to SD4, etc. installed in a centrifugal compressor train 200. Fig. 3 shows an example of measurement values ​​in the condition monitoring diagnosis of this embodiment. Measurement values ​​at each time are input, and from these measurement values, reference data that serves as a standard and diagnosis data that is the target of diagnosis are obtained.

[0018] The reference data generation unit 3 extracts reference data from the data for a period considered normal among the measurement values ​​input to the measurement value input unit 2. The reference data is standard data used by the abnormality degree calculation unit 5 to calculate the degree to which the measurement values ​​for the period to be diagnosed deviate from a standard state considered normal as the degree of abnormality. As the degree of abnormality increases, the possibility increases that normal function is impaired or has been impaired. The reference data generated by the reference data generation unit 3 is stored in the reference data storage unit 4.

[0019] The abnormality degree calculation unit 5 acquires reference data, which is a measurement value of the centrifugal compressor 201 under normal conditions, from the reference data storage unit 4, and calculates the degree of abnormality at any time using this reference data as a reference. As a calculation process of the abnormality degree calculation unit 5, for example, a method of calculating Kullback-Leibler divergence based on the correlation between parameters can be used. In other words, by using the Kullback-Leibler divergence, it is possible to calculate the degree of abnormality taking into account the correlation between parameters. However, this is not limited to this. The calculated degree of abnormality is output to the abnormality cause estimation unit 7 as the degree of abnormality of the centrifugal compressor.

[0020] The reference data generation unit 3 will be further explained using the drawings. FIG. 4 shows an example of one parameter of the first reference data for the condition monitoring diagnosis of this embodiment. The measurement values ​​shown in FIG. 3 are graphed with time on the horizontal axis. The first reference data is data that includes a period considered to be normal among the measurement values ​​input to the measurement value input unit 2, and is a first time period A that includes a period that serves as reference data that is used as a basis when the abnormality degree calculation unit 5 calculates the abnormality degree. The first time period A is specified by the person implementing the diagnosis.

[0021] As can be seen from the change in rotation speed obtained from the centrifugal compressor shaft vibration sensor SC7, the first time period A in Figure 4 includes the transient state of the entire process, which includes startup (SU), temporary shutdown, and restart (SS). Furthermore, the bearing temperature obtained from the centrifugal compressor bearing temperature sensor SC5 shows a momentary temperature rise (TU). This temperature rise is extremely rapid and large compared to typical temperature changes. Furthermore, the temperature subsequently returns to its original value, indicating measurement noise. If the anomaly degree calculation unit 5 were to use the first reference data, which contains such transient states and momentary noise, as reference data to calculate the anomaly degree, the transient states and noise would be considered normal, potentially reducing the accuracy of the calculated anomaly degree. Therefore, the reference data generation unit 3 generates reference data by removing the transient states and momentary noise from the first reference data.

[0022] FIG. 5 is a diagram showing the configuration of the reference data generation unit for the condition monitoring and diagnosis of this embodiment. FIG. 6 is a diagram showing examples of first reference data and second time periods, the abnormality level for each second time period, and one parameter value for each second time period, and is also a diagram showing examples of first time period A, second time period B, the abnormality level K for each second time period, and one parameter value P in the reference data generation unit 3. The reference data generation unit 3 divides the first time period A into multiple second time periods B included in the first time period A, and calculates the abnormality level for the second time period B using the measurement data for the first time period A (i.e., the first reference data) as reference data. To calculate the Kullback-Leibler information based on the correlation between parameters as the abnormality level, a group of measurement data is required to capture the correlation. Therefore, the second time period is not a single time point but a time period including measurement data from multiple times.

[0023] Because the first time period A contains transient states and noise, if the first reference data is used as is to calculate an anomaly, the behavior of the measurement data during the transient states and noise may be deemed normal, potentially reducing the accuracy of the calculation of the degree of anomaly. Therefore, during diagnosis, it is desirable to calculate the degree of anomaly using reference data from which the transient states and noise have been removed. However, because the behavior of the measurement data during transient states and noise differs from the behavior of the entire first reference data, an appropriate threshold can be set to extract the second time period BN containing the transient states and noise. Therefore, in this embodiment, the degree of anomaly of the measurement data during the second time period B relative to the first reference data is calculated. If the degree of anomaly exceeds a specified first threshold T1, it is determined that the measurement data contains a transient state or noise, and the second time period BN is extracted. The first threshold T1 may be set, for example, using a standard deviation.

[0024] In this case, it is possible to generate reference data by removing all of the extracted second time zone BN. However, the duration of noise, which is an instantaneous fluctuation, is short and its proportion to the second time zone BN is small. Therefore, removing all of the extracted second time zone BN would also remove measurement data that is suitable as reference data and is included in the same second time zone BN as the noise. Since the accuracy of calculating the degree of abnormality during diagnosis improves with the amount of reference data, in plant equipment where it is difficult to obtain measurement data that can be considered normal, removing the entire second time zone BN containing noise may reduce the calculation accuracy.

[0025] Therefore, if one parameter value P of the measurement values ​​exceeds the second threshold value T2 during the extracted second time period BN, the time t at which the one parameter value P exceeds the second threshold value T2 is removed as inappropriate data. The one parameter value P may be, for example, a bearing temperature or a discharge pressure, and is a parameter value that does not change rapidly with each measurement time interval during normal operation. While the example illustrates the use of the measured value as the one parameter value, this is not limiting. For example, a value obtained by differentiating the measured value may be used. For example, the second threshold value T2 may be a reference value for a predetermined parameter. In this embodiment, the threshold determination for the abnormality level during the second time period B and the one parameter value P using the first and second threshold values ​​is performed once each, but similar processing may be performed multiple times for both or only one of them. This improves the reliability of the reference data. Furthermore, by providing a threshold value input unit for inputting at least one of the first threshold value T1 and the second threshold value T2, appropriate values ​​can be input, thereby creating appropriate reference data.

[0026] As described above, when the amount of data change exceeds the first threshold value T1 during a second time period B included in a first time period A of the measurement data of the state of the plant equipment, and when one parameter value exceeds the second threshold value T2 during the second time period B, the data for the time period during which the one parameter value exceeded the second threshold value is removed from the measurement data to create reference data, and the measurement data for monitoring the state of the plant equipment is compared with the reference data to monitor the state of the equipment, and the comparison with the noise-removed reference data enables appropriate abnormality determination. Furthermore, by repeating this noise removal, the reliability of the reference data is improved, enabling more appropriate abnormality determination.

[0027] Furthermore, noise in the reference data used for anomaly determination can be removed, and anomalies in the plant equipment can be diagnosed with high accuracy. In other words, it is possible to provide a condition monitoring and diagnostic method, a condition monitoring and diagnostic device, and a condition monitoring and diagnostic system that can easily and accurately diagnose anomalies in plant equipment that has little measurement data that can be used as reference data.

[0028] 7 shows an example of reference data for the condition monitoring and diagnosis of this embodiment, and is a diagram showing an example of reference data generated by the reference data generating unit 3. The reference data generating unit 3 removes time periods including transient states and noise.

[0029] Here, by using, for example, any one of the bearing temperature, rotation speed, and flow rate of the centrifugal compressor, which is the target plant, as one parameter value, the state of the centrifugal compressor can be appropriately monitored.

[0030] The anomaly cause estimation unit 7 has a function of estimating the cause of an anomaly based on the anomaly degree calculation result received from the anomaly degree calculation unit 5. The anomaly cause estimation unit 7 constructs an anomaly cause estimation model 6 in advance in accordance with an abnormal event that may occur in the configuration of the centrifugal compressor train 200. Note that, as the anomaly cause estimation model 6, for example, a Bayesian network in which causal relationships are organized in a directed acyclic graph structure is used, but the present invention is not limited to this.

[0031] The result output unit 8 outputs the degree of abnormality calculated by the degree of abnormality calculation unit 5 and the cause of abnormality estimated by the abnormality cause estimation unit 7 as a diagnostic result in the form of a graph or a table.

[0032] The present invention is not limited to the above-described embodiments and includes various modifications. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and the present invention is not necessarily limited to those including all of the described configurations. Furthermore, some or all of the above-described configurations, functions, etc. may be realized by designing, for example, an integrated circuit. Furthermore, the above-described configurations, functions, etc. may be realized by software in which a processor interprets and executes a program that realizes each function.

[0033] According to this embodiment, in a condition monitoring and diagnostic method for measuring measurement data including multiple parameter values ​​obtained from multiple measuring instruments of plant equipment, measurement data during a period when the equipment is functioning normally is used as reference data, measurement data during a period to be diagnosed is used as diagnostic data, and the amount of change (degree of abnormality) in the diagnostic data from the reference data is calculated. For measurement data during a first time period during which the equipment is functioning relatively normally, which is a candidate for reference data, the measurement data during the first time period is used as first reference data to calculate the degree of abnormality of the measurement data during a second time period included in the first time period. The degree of abnormality increases when the behavior of the measurement data during the second time period differs from that of the first reference data. When the degree of abnormality exceeds a first threshold and one parameter value of the measurement data during the second time period exceeds a second threshold, the measurement data during the time when the one parameter value exceeded the second threshold is excluded from the second time period and used as reference data for diagnosis.

[0034] As described above, it is possible to generate reference data from the first reference data by excluding only time periods that have a high degree of abnormality compared to the first reference data as a whole and in which one parameter value is large, thereby minimizing the amount of measurement data that cannot be used as reference data. [Explanation of symbols]

[0035] 1... Condition monitoring and diagnosis system, 2... Measurement value input unit, 3... Reference data generation unit, 4...Reference data storage unit, 5...Abnormality degree calculation unit, 6...Abnormality cause estimation model, 7...abnormality cause estimation unit, 8...result output unit, 200... centrifugal compressor train, 201... centrifugal compressor, 202... driver, 203...coupling, 204...centrifugal compressor rotor, 205...driver rotor, SC1 to SC4: Centrifugal compressor shaft vibration sensors, SC5~SC6...Centrifugal compressor bearing temperature sensor, SC7...rotation speed sensor, SC8...suction pressure sensor, SC9...discharge pressure sensor, SC10...flow sensor, SD1 to SD4: Drive shaft vibration sensors, SD5~SD6...Driver bearing temperature sensor, A...First time period, B...Second time period, BN: a second period including transients and noise; K...anomaly level for each second time period, P...one parameter value, SU: Start-up, SS: Temporary stop and restart, TU: Instantaneous bearing temperature rise, T1...first threshold, T2...second threshold

Claims

1. A plant equipment status monitoring method for monitoring a status of plant equipment, comprising: a reference data generating step of generating reference data by excluding data for a time when a data change amount exceeds a first threshold value in a second time period included in a first time period of the measurement data of the state of the plant equipment and one parameter value exceeds a second threshold value in the second time period; A plant equipment status monitoring method comprising the step of comparing the measurement data for monitoring the status of the plant equipment with the reference data to monitor the status of the equipment.

2. 2. The plant equipment status monitoring method according to claim 1, A plant equipment status monitoring method, characterized in that the reference data generating step is performed at least once for the entire range or a part of the created reference data.

3. 3. The plant equipment status monitoring method according to claim 1, further comprising: A method for monitoring the status of plant equipment, characterized in that, when a data change amount exceeds the first threshold value during the second time period of the measurement data of the status of the plant equipment for the entire range or a part of the created reference data, the second time period is excluded from the measurement data to create the reference data.

4. 3. The plant equipment status monitoring method according to claim 1, further comprising: A plant equipment status monitoring method, characterized in that the amount of change in data obtained by comparing the monitoring measurement data of the plant equipment with the reference data is defined as Kullback-Leibler divergence based on the correlation between parameters.

5. 3. The plant equipment status monitoring method according to claim 1, further comprising:

10. A plant equipment status monitoring method, wherein the one parameter value is one of a bearing temperature, a rotation speed, and a flow rate of a centrifugal compressor.

6. In a plant equipment status monitoring system that monitors the status of plant equipment, a reference data generating unit that, when a data change amount exceeds a first threshold value in a second time period included in a first time period of the measurement data of the state of the plant equipment and one parameter value exceeds a second threshold value in the second time period, generates reference data by excluding data for a time period in which the one parameter value exceeds the second threshold value from the measurement data; A plant equipment status monitoring system comprising an abnormality degree calculation unit that compares the measurement data for monitoring the status of the plant equipment with the reference data to calculate an abnormality degree.

7. 7. The plant equipment status monitoring system according to claim 6, A plant equipment status monitoring system comprising an abnormality cause estimation unit that estimates the cause of an abnormal event from a pre-stored abnormality cause estimation model and the calculation result of the abnormality degree calculation unit.

8. 8. The plant equipment status monitoring system according to claim 7, a result output unit that outputs the degree of abnormality calculated by the degree of abnormality calculation unit and the cause of the abnormal event estimated by the abnormality cause estimation unit as diagnostic results.

9. 7. The plant equipment status monitoring system according to claim 6, A plant equipment status monitoring system comprising a threshold value input unit for inputting at least one of the first threshold value and the second threshold value.

Images