System and method for estimating the cause of device abnormality

Abstract

To provide an abnormality cause estimation method and an abnormality cause estimation system that simply and accurately estimate the cause of an abnormality in various devices such as plant equipment.SOLUTION: An abnormality cause estimating system 1 includes: a measurement value input unit 2 that receives measurement data regarding the operation of a device when an abnormality occurs; an abnormality degree calculation unit 3 that calculates an abnormality degree distribution of the device from the received measurement data; and an abnormality cause estimation unit 5 that weights the calculated abnormality degree distribution by the occurrence frequency distribution of parameter abnormalities specified by the characteristics of the device, and estimates the cause of the abnormality that has occurred in the device.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a technique for estimating the cause of an abnormality (cause of an abnormality) in a plant, particularly in equipment such as a compressor and other plant equipment. Note that the term "equipment" as used herein includes various facilities, products, appliances, machines, and other items that perform a predetermined operation and thereby perform a desired function. [Background technology]

[0002] For the purpose of stable operation and streamlining of maintenance work, technologies for monitoring and diagnosing various equipment in chemical plants and power plants are being introduced. When an abnormality is detected in plant equipment during such monitoring, it is expected that operation can be continued and maintenance work time can be reduced by detecting it early and taking measures. In order to take measures more effectively, it is important to have diagnostic technology that can estimate the cause of an abnormality when an abnormality is detected.

[0003] As an example of a diagnostic technique, for example, Patent Document 1 describes a technique in which the degree of change of each attribute of diagnostic object data containing multiple attributes measured from the diagnostic object is calculated, a diagnostic object change degree pattern is generated which is a combination of the degrees of change of each attribute, and pattern matching is performed between the diagnostic object change degree pattern and a reference change degree pattern in a reference change degree pattern storage device to diagnose the diagnostic object (see abstract).

[0004] Furthermore, for example, Patent Document 2 discloses a technique for generating additional conditions and a case set so as to obtain a probability distribution as a new case set when a new diagnosis situation not in the case database is input in a fault diagnosis method that estimates the cause of a fault in the current situation from the frequency distribution of the fault causes in a case database. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] International Publication No. 2008 / 114863 [Patent Document 2] Japanese Patent Application Publication No. 5-35484 Summary of the Invention [Problem to be solved by the invention]

[0006] For large rotating machines such as compressors used in various plants, the configuration, connections, and measurement parameters of the equipment vary from plant to plant, and the number of machines that can be compared under the same conditions is limited. In addition, large rotating machines such as compressors are required to have high reliability because they affect the productivity of the plant. Therefore, the frequency of failure is relatively low, and there are cases where the data at the time of abnormality is insufficient to link the measurement parameters and the cause of the abnormality using machine learning or other methods from the data at the time of actual abnormality occurrence. For this reason, it is sometimes difficult to directly compare the details of the parameter behavior patterns at the time of abnormality and estimate the cause of the abnormality. In this case, instead of directly estimating the cause of the abnormality from the pattern of parameter behavior, it is necessary to determine whether or not each parameter is abnormal, and then estimate the cause of the abnormality from the abnormality presence / absence pattern, which can be time-consuming.

[0007] An object of the present invention is to simply and accurately estimate the cause of an abnormality in equipment such as plant equipment. [Means for solving the problem]

[0008] In order to solve the above problems, in the present invention, an anomaly degree distribution is calculated based on measurement data indicating the measurement values ​​of a target device, and the anomaly degree distribution is weighted by an occurrence frequency distribution specified by the characteristics of the device, thereby inferring the cause of an anomaly that has occurred in the device.

[0009] More specifically, the equipment anomaly cause estimation device has a measurement value input unit that receives measurement data regarding the operation of the equipment when an abnormality occurs, an anomaly degree calculation unit that calculates an anomaly degree distribution of the equipment from the received measurement data, and an anomaly cause estimation unit that weights the calculated anomaly degree distribution by a parameter anomaly occurrence frequency distribution specified by the characteristics of the equipment, and estimates the cause of an abnormality that has occurred in the equipment.

[0010] The present invention also includes an equipment abnormality cause estimating method using the equipment abnormality cause estimating device, a computer program for causing the equipment abnormality cause estimating device to function as a computer, and a storage medium for storing the same. Effect of the Invention

[0011] According to the present invention, it is possible to easily estimate the cause of an abnormality in an apparatus with high accuracy. [Brief description of the drawings]

[0012] [Figure 1] 1 is a functional block diagram of an abnormality cause estimating system 1 according to an embodiment of the present invention. [Diagram 2] FIG. 1 is a configuration diagram of a centrifugal compressor train 200 to be managed in one embodiment of the present invention. [Diagram 3] FIG. 13 is a diagram showing measurement data (normal state measurement data 4) used in one embodiment of the present invention. [Figure 4] FIG. 2 is a diagram showing an anomaly cause estimation model 6 constructed in one embodiment of the present invention. [Diagram 5] 1 is a diagram showing an example of an abnormality degree distribution used in an abnormality cause estimation system 1 according to one embodiment of the present invention. [Figure 6] 1 is a diagram showing an example of occurrence frequency distribution in an anomaly cause estimation model 6 in an embodiment of an anomaly cause estimation system 1 according to one embodiment of the present invention. FIG. [Figure 7] 3A to 3C are diagrams showing examples of abnormality cause estimation results in the embodiment of the abnormality cause estimation system 1 of the present invention. [Figure 8] 1 is a diagram showing an implementation example of an anomaly cause estimation system 1 in one embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] An embodiment of the present invention will be described in detail below with reference to the drawings as appropriate. In addition, the same reference numerals are used for equivalent components throughout the drawings. FIG. 1 is a functional block diagram of an anomaly cause estimation system 1 in this embodiment. The anomaly cause estimation system 1 corresponds to the equipment anomaly cause estimation device of the present invention, and estimates the cause of anomaly in equipment such as plant equipment. For this purpose, in FIG. 1, the anomaly cause estimation system 1 has a measurement value input unit 2, an anomaly degree calculation unit 3, an anomaly cause estimation unit 5, and a result output unit 7, and stores normal measurement data 4 and an anomaly cause estimation model 6 in a storage unit. Note that the storage unit may be configured separately from the anomaly cause estimation system 1, or the normal measurement data 4 and the anomaly cause estimation model 6 may be stored in separate storage units. In addition, the anomaly cause estimation system 1 can be realized by a computer. An implementation example of this will be described later with reference to FIG. 8. The functions of each unit will be described below.

[0014] First, the measurement value input unit 2 receives measurement data from a sensor that detects information related to the operation of the equipment, including its operation. In this embodiment, the equipment is a centrifugal compressor train 200 of centrifugal compressors, which is a representative plant equipment. The sensor is provided in the plant equipment. Preferably, the plant equipment is provided with a plurality of sensors. Here, the information related to the operation and the measurement data indicate the shaft vibration and rotation speed of the plant equipment. Then, the sensor digitizes the information related to the operation into a signal, and the measurement data is identified. The measurement data can also be expressed as a measurement value.

[0015] The configuration of the plant equipment will be described below by taking a centrifugal compressor train 200 as an example. Fig. 2 is a configuration diagram of the centrifugal compressor train 200 that is a managed object in this embodiment. In Fig. 2, the centrifugal compressor train 200 includes a centrifugal compressor 201 and a driver 202 connected via a coupling 203.

[0016] Also, in FIG. 2, the centrifugal compressor 201 is provided with four axial vibration sensors SC1 to SC4 for measuring the axial vibration of the centrifugal compressor rotor 204, a pressure sensor SC5 for measuring the suction pressure, a pressure sensor SC6 for measuring the discharge pressure, and a sensor SC7 for measuring the rotation speed.

[0017] The driving machine 202 is also provided with sensors such as four axial vibration sensors SD1 to SD4 that measure axial vibration of the driving machine rotor 205. These sensors are merely examples, and other sensors may be provided, or at least some of them may be omitted.

[0018] Further, the measurement value input unit 2 of the abnormality cause estimation system 1 in FIG. 1 receives measurement data such as axial vibration, suction pressure, discharge pressure, and rotation speed obtained from the sensors SC1 to SC7, SD1 to SD4, etc. installed in the centrifugal compressor train 200. Here, these measurement data will be described. FIG. 3 is a diagram showing measurement data used in this embodiment. In FIG. 3, normal measurement data 4 measured when the centrifugal compressor train 200 is operating normally is shown as an example, but measurement data other than normal has the same structure (items). As shown in FIG. 3, each measurement data such as axial vibration (SC1, SC2), suction pressure (SC5), discharge pressure (SC6), rotation speed (SC7), axial vibration (SD1), etc. is measured periodically (at each time). The measurement time is preferably every period, particularly every certain period, but is not limited to this. The normal measurement data 4 in FIG. 1 has the same structure as FIG. 3, and is data measured when the centrifugal compressor train 200 is operating normally (at normal times). The normal state indicates that the managed object, such as the centrifugal compressor train 200, is in a predetermined state, for example, a state in which the managed object operates while exerting a desired performance. The normal state is the basis for diagnosing the managed object.

[0019] Returning to Fig. 1, the description of each part of the anomaly cause estimation system 1 will be continued. The anomaly degree calculation part 3 calculates the anomaly degree of each measurement parameter by using the measurement data received by the measurement value input part 2. That is, the measurement value input part 2 receives measurement data when the centrifugal compressor train 200, which is an equipment, is abnormal. The measurement value input part 2 can also receive measurement data when the centrifugal compressor train 200 is normal. Here, the anomaly degree of the measurement parameter represents the degree to which the measurement data deviates from the normal state, and a large anomaly degree indicates that normal function is impaired or there is a high possibility that it is impaired.

[0020] Moreover, the anomaly degree calculation unit 3 acquires normal state measurement data 4, which is measurement data of the centrifugal compressor train 200 under normal conditions, from the storage unit, and calculates the degree of anomaly of each measurement parameter at a certain time based on the normal state measurement data 4. For example, the Mahalanobis-Taguchi method is used as the process of the anomaly degree calculation unit 3, but is not limited thereto. Moreover, the anomaly degree calculation unit 3 outputs a combination of the calculated degrees of anomaly of the multiple measurement parameters to the anomaly cause estimation unit 5 as an anomaly degree distribution of the centrifugal compressor train 200.

[0021] Moreover, the anomaly cause estimation unit 5 estimates the cause of the anomaly based on the anomaly degree distribution calculated by the anomaly degree calculation unit 3. Moreover, the anomaly cause estimation unit 5 constructs an anomaly cause estimation model 6 in advance according to an abnormal event that may occur in the configuration of the centrifugal compressor train 200. Note that, in this embodiment, the anomaly cause estimation model 6 is described using a Bayesian network in which causal relationships are organized in a directed acyclic graph structure, but is not limited thereto.

[0022] Moreover, the anomaly cause estimation model 6 is a model showing a causal relationship between an abnormal event and a measurement parameter that causes an abnormality when the abnormal event occurs. As described above, the anomaly cause estimation model 6 is constructed by the anomaly cause estimation unit 5 using a Bayesian network. Here, Fig. 4 is a diagram showing the anomaly cause estimation model 6 constructed in this embodiment.

[0023] The construction of the anomaly cause estimation model 6 will be described below. The anomaly cause estimation unit 5 sets a conditional probability P(B|A) that a child event B occurs when a parent event A occurs. With this, the anomaly cause estimation unit 5 calculates, by (Equation 1), how frequently an anomaly occurs in each measurement parameter when a certain anomaly cause event occurs.

[0024] Here, P 0 is the probability (prior probability) of an abnormal cause event occurring in the Bayesian network, and P 1 , P 2 , … P n is the conditional probability set for each link. (Equation 1) represents the probability (occurrence frequency) that an abnormality will occur in a specific parameter when a certain abnormality-causing event occurs. The prior probability and the conditional probability are set based on design information, past cases, etc. P=P 0 ×P 1 ×P 2 ×…×P n (Number 1) As described above, when a certain abnormality cause event occurs, a combination of occurrence frequencies of each parameter abnormality is obtained for each abnormality cause, and a distribution of occurrence frequencies of parameter abnormalities corresponding to each abnormality cause (parameter abnormality occurrence frequency distribution) is obtained. This parameter abnormality occurrence frequency distribution can be specified by the characteristics of a device such as the centrifugal compressor train 200. The characteristics include the specifications, design, etc. of the device.

[0025] FIG. 5 shows an example of an anomaly degree distribution received from the anomaly degree calculation unit 3, and FIG. 6 shows an example of a parameter anomaly occurrence frequency distribution by anomaly cause obtained from the anomaly cause estimation model 6 in this embodiment. In each figure, the measurement parameters are expressed as Para1 to Para10. These are associated with the anomaly degree of the measurement data obtained from each sensor and the (anomaly) occurrence frequency of each sensor. Here, the anomaly degree distribution is calculated based on the measurement value at a certain time, and one anomaly degree distribution is obtained for one set of measurement data consisting of multiple measurement data. On the other hand, there are as many parameter anomaly occurrence frequency distributions as the number of anomaly causes assumed in the anomaly cause estimation model 6. In FIG. 6, five cases of anomaly causes A to E are illustrated.

[0026] Furthermore, the anomaly cause estimation unit 5 weights the anomaly distribution by the parameter anomaly occurrence frequency distribution. This allows the anomaly cause estimation unit 5 to estimate the cause of the anomaly in the centrifugal compressor train 200. At this time, the anomaly cause estimation unit 5 sets the anomaly cause of the parameter anomaly occurrence frequency distribution according to the anomaly distribution calculated by the anomaly calculation unit 3 as the estimation result. For example, the anomaly cause estimation unit 5 estimates the anomaly cause that has the parameter anomaly occurrence frequency distribution most similar to the anomaly distribution. At this time, the anomaly cause estimation unit 5 sets the anomaly cause of the parameter anomaly occurrence frequency distribution with a higher order, that is, with higher priority, as the estimation result, as the anomaly cause becomes more similar to the anomaly distribution (the higher the similarity). Here, the estimation result includes setting a predetermined number of anomaly causes from the top and setting those with a similarity equal to or higher than a predetermined value as the anomaly cause.

[0027] Here, as a weighting method, the abnormality degree distribution is defined as a first vector α, and the parameter abnormality occurrence frequency distribution is defined as a second vector β k Then, the inner product of the two, γ k =α β k Calculate the inner product γ kIt is determined that the greater the value of , the higher the similarity between the anomaly degree distribution and the parameter anomaly occurrence frequency distribution, and it is estimated that the possibility of the anomaly cause corresponding to the parameter anomaly occurrence frequency distribution is high. In order to correctly perform weighting using the inner product, the anomaly cause estimation unit 5 arranges the order of the parameters in the anomaly degree distribution and the order of the parameters in the parameter anomaly occurrence frequency distribution to be the same. By using the inner product, it is possible to easily check the similarity between both distributions. In this embodiment, the inner product has been used as the weighting method, but the method is not limited to this.

[0028] The inner product γ calculated from the anomaly distribution in Fig. 5 and the parameter anomaly occurrence frequency distribution in Fig. 6 k An example of the results of the anomaly cause estimation is shown in Fig. 7. The inner product γ k The larger the value, the higher the priority of the abnormality cause event that is occurring.

[0029] Furthermore, the result output unit 7 receives the anomaly degree distribution of the measurement values ​​and the estimated anomaly cause events from the anomaly cause estimation unit 5, and outputs them as diagnostic results in the form of graphs or tables. Therefore, the result output unit 7 can be realized by a display device such as a display. Furthermore, the result output unit 7 can be realized as an interface device for connecting to a terminal device or a display device.

[0030] The above is the description of the functions of the anomaly cause estimation system 1 in this embodiment, and an implementation example thereof will be described below. FIG. 8 is a diagram showing an implementation example of the anomaly cause estimation system 1 in this embodiment. In FIG. 8, the centrifugal compressor trains 200 in a plurality of plants A1000-1 and B1000-1 are subject to management. The plant A1000-1 is provided with a plurality of centrifugal compressor trains 200-1 (equipment A) and 200-2 (equipment B). Furthermore, the plant A1000-1 is provided with terminal devices 500-1 and 500-2 for monitoring the centrifugal compressor trains 200-1 (equipment A) and 200-2. Note that the numbers of the plants 100, the centrifugal compressor trains 200, and the terminal devices 500 are merely examples, and are not limited to the example in FIG. 8. The terminal devices 500-1 and 500-2 can be realized by a PC or a tablet.

[0031] Further, the centrifugal compressor trains 200-1 and 200-2 are connected to control devices 400-1 and 400-2, respectively, to control the operation of the centrifugal compressor trains 200-1 and 200-2. The control devices 400-1 and 400-2 and the terminal devices 500-1 and 500-2 are connected to a local network in the plant A1000-1, such as an intranet. As a result, the terminal devices 500-1 and 500-2 grasp the operating conditions, including abnormalities, of the centrifugal compressor trains 200-1 and 200-2, and output control instructions to the control devices 400-1 and 400-2. Furthermore, the centrifugal compressor trains 200-1 and 200-2 are provided with sensors, as described above. However, the number and types of sensors may be different for each of the centrifugal compressor trains 200-1 and 200-2. The managed objects may be other devices than the centrifugal compressor trains.

[0032] The local network of plant A 1000-1 is connected to the local network of plant B 1000-2 and the anomaly cause estimation system 1 via a wide area network 300 such as the Internet. However, the anomaly cause estimation system 1 may be provided on the local networks of plant A 1000-1 and plant B 1000-2. The above-mentioned sensors are also connected to the local networks. As a result, the measurement data of the sensors is notified to the anomaly cause estimation system 1 via the local network and the wide area network 300.

[0033] Next, a description will be given of the configuration of the anomaly cause estimating system 1. The anomaly cause estimating system 1 can be realized by a computer such as a server. An implementation example will be described below with reference to FIG.

[0034] In FIG. 8, the anomaly cause estimating system 1 includes a CPU 101, a RAM 102, a ROM 103, a communication device 104, and a storage device 105, which are connected to each other via a communication path.

[0035] First, the CPU (Central Processing Unit) 101 can be realized by a processor, and executes calculations according to the anomaly degree calculation program 106, the anomaly cause estimation program 107, and the equipment operation program 108 stored in the storage device 105. The RAM (Random Access Memory) 102 is used as a so-called main storage device, and the anomaly degree calculation program 106, the anomaly cause estimation program 107, and the equipment operation program 108, as well as data and information for calculating these, are deployed. The ROM (Read Only Memory) 108 stores firmware and the like of the anomaly cause estimation system 1. The communication device 104 has an interface function with other devices. That is, the communication device 104 is connected to the wide area network 300, and has the functions of the measurement value input unit 2 and the result output unit 7 in FIG. 1. The result output unit 7 may be realized by the terminal device 500-1 or 500-2, or a display device may be provided in the anomaly cause estimation system 1.

[0036] The storage device 105 corresponds to the above-mentioned storage unit, and stores the anomaly cause estimation program 107, the equipment operation program 108, the normal state measurement data 4, and the anomaly cause estimation model 6. This storage device 105 can be realized by various storage media such as an external HDD (Hard Disk Drive), SSD (Solid State Drive), or memory card. Furthermore, the storage device 105 may be realized by a device separate from the anomaly cause estimation system 1, which is connected to the wide area network 300 like a file server.

[0037] Here, the abnormality degree calculation program 106 is a program for realizing the function of the abnormality degree calculation unit 3. Moreover, the abnormality cause estimation program 107 is a program for realizing the function of the abnormality cause estimation unit 5. Furthermore, the equipment operation program 108 creates control commands for controlling the operation of the managed equipment such as the centrifugal compressor trains 200-1 and 200-2. Note that the equipment operation program 108 may be omitted or may be provided in the terminal devices 500-1 and 500-2.

[0038] As described above, the anomaly cause estimating system 1 can be constructed as a so-called cloud system that can comprehensively manage multiple plants. However, it may also be realized as a local system specialized for a plant, such as an in-house system. Furthermore, the anomaly cause estimating system 1 may be realized by a computer called a standalone computer.

[0039] The present invention is not limited to the above-described embodiment, and various modified examples are included. For example, the above-described embodiment has been described in detail to clearly explain the present invention, and is not necessarily limited to the embodiment having all of the described configurations. In addition, the above-described configurations, functions, etc. may be realized by designing a part or all of them, for example, in an integrated circuit. In addition, the equipment to be managed is not limited to manufacturing equipment such as the centrifugal compressor train 200, but also includes other industrial equipment such as construction machinery and agricultural machinery, and consumer equipment.

[0040] According to one example of this embodiment, the degree of anomaly (amount of change from a reference) is obtained from the measurement data obtained from a plurality of measuring instruments of the centrifugal compressor train 200 plant equipment, and an anomaly distribution consisting of the plurality of degrees of anomaly is obtained. Meanwhile, an anomaly cause estimation model is constructed in which the target anomaly cause is previously linked to an influence parameter that is assumed to be affected when the anomaly cause event occurs, and the occurrence frequency distribution of each parameter anomaly for each anomaly cause is obtained. By weighting the anomaly distribution obtained from the measurement data with the occurrence frequency distribution obtained from the anomaly cause estimation model, similar anomaly causes are estimated with a high priority, and it becomes possible to estimate the anomaly cause simply and with high accuracy without directly comparing the anomaly distribution at the time of an anomaly occurrence. Also, in one example of this embodiment, the estimated anomaly causes may be ranked without determining the presence or absence of an anomaly. [Explanation of symbols]

[0041] 1. Anomaly cause estimation system 2. Measurement value input section 3... Abnormality calculation section 4. Normal measurement value database 5... Abnormal cause estimation section 6. Anomaly cause estimation model 7. Result output section 200···Centrifugal Compressor Train 201...Centrifugal compressor 202 Drive unit 203 Coupling 204...Centrifugal compressor rotor 205 Drive rotor SC1~SC4: Centrifugal compressor shaft vibration sensor SC5···Suction pressure sensor SC6···Discharge pressure sensor SC7···Rotational Speed ​​Sensor SD1~SD4: Drive shaft vibration sensor SE1~SE4...Shaft vibration sensors for other rotating machines

Claims

1. a measurement value input unit that receives measurement data regarding an operation of the device when an abnormality occurs; an anomaly degree calculation unit that calculates an anomaly degree distribution of the device from the received measurement data; An equipment anomaly cause estimation device having an anomaly cause estimation unit that weights the calculated anomaly degree distribution by a parameter anomaly occurrence frequency distribution specified by characteristics of the equipment, and estimates the cause of an anomaly that has occurred in the equipment.

2. 2. The device for estimating a cause of an equipment abnormality according to claim 1, The anomaly cause estimation unit is an equipment anomaly cause estimation device that obtains the parameter anomaly occurrence frequency distribution by calculating a combination of occurrence frequencies of each parameter anomaly for each anomaly cause.

3. 3. The device for estimating a cause of an equipment abnormality according to claim 2, The anomaly cause estimation unit is an equipment anomaly cause estimation device that estimates, as the anomaly cause, an anomaly cause that has a parameter anomaly occurrence frequency distribution that is most similar to the anomaly degree distribution.

4. 2. The device for estimating a cause of an equipment abnormality according to claim 1, The weighting is performed by calculating the inner product of the first vector representing the anomaly degree distribution and the second vector representing the distribution of the frequency of occurrence of each parameter anomaly by the anomaly cause.

5. 3. The device for estimating a cause of an equipment abnormality according to claim 2, The anomaly cause estimation unit is an equipment anomaly cause estimation device that obtains the parameter anomaly occurrence frequency distribution by using a Bayesian network that organizes physical causal relationships.

6. 3. The device for estimating a cause of an equipment abnormality according to claim 2, The anomaly cause estimation unit aligns the order of parameters in the anomaly degree distribution and the parameter anomaly occurrence frequency distribution.

7. In a method for estimating the cause of an equipment abnormality by an equipment abnormality cause estimation device, The measurement value input unit receives measurement data on the operation of the equipment when an abnormality occurs, an anomaly degree calculation unit calculates an anomaly degree distribution of the device from the received measurement data; An equipment anomaly cause estimating method in which an anomaly cause estimation unit weights the calculated anomaly degree distribution by a parameter anomaly occurrence frequency distribution specified by characteristics of the equipment, and estimates the cause of an anomaly that has occurred in the equipment.

8. The method for estimating a cause of an equipment abnormality according to claim 7, The method for estimating the cause of an abnormality in an equipment, wherein the abnormality cause estimation unit determines the parameter abnormality occurrence frequency distribution by determining a combination of the occurrence frequency of each parameter abnormality for each abnormality cause.

9. The method for estimating a cause of an equipment abnormality according to claim 8, The method for estimating the cause of an abnormality in an equipment, in which the abnormality cause estimation unit estimates, as the cause of the abnormality, an abnormality cause that has a parameter abnormality occurrence frequency distribution that is most similar to the abnormality degree distribution.

10. The method for estimating a cause of an equipment abnormality according to claim 7, The weighting is performed by calculating the inner product of the anomaly degree distribution as a first vector and the occurrence frequency distribution of each parameter anomaly by the anomaly cause as a second vector.

11. The method for estimating a cause of an equipment abnormality according to claim 8, The method for estimating the cause of an abnormality in an equipment, wherein the abnormality cause estimation unit determines the parameter abnormality occurrence frequency distribution using a Bayesian network that organizes physical causal relationships.

12. The method for estimating a cause of an equipment abnormality according to claim 8, The method for estimating the cause of an abnormality in an equipment, wherein the abnormality cause estimation unit aligns the order of parameters in the abnormality degree distribution and the parameter abnormality occurrence frequency distribution.

Images