Diagnostic device, diagnostic method, and diagnostic system

Abstract

To provide a diagnostic device which estimates a cause of abnormality while discriminating an abnormality and a transition state.SOLUTION: A diagnostic device includes: a measurement value input unit for inputting each measurement value obtained from a plurality of sensors provided in an object device; an abnormality determination unit for determining presence or absence of each measurement value; and an abnormality cause estimation unit for estimating an abnormality cause of the object device based on each measurement value determined for abnormality by the abnormality determination unit. The abnormality cause estimation unit estimates the abnormality cause of the object device by using a combination of the presence or absence of each measurement value obtained from the abnormality determination unit, and an abnormality cause database in which the combination of the measurement values where abnormalities are detected, and abnormality causes containing that the object device is in a transition state are stored in association with each other.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a diagnosis device or the like that estimates the cause of an abnormality in fluid machinery or the like. [Background technology]

[0002] Chemical plants and power plants are increasingly introducing state monitoring and diagnostic technologies for plant equipment to ensure stable operation and streamline maintenance work. Early detection and countermeasures for abnormalities in plant equipment can be expected to enable continued operation and shorten maintenance work time. In order to implement countermeasures more effectively, diagnostic technologies that can estimate the cause of an abnormality when it occurs are important.

[0003] As an example of a diagnostic technique, for example, Patent Document 1 states that "a phenomenon pattern extraction unit 104 extracts phenomenon patterns of past sensor signals of the equipment. A related information linking unit 105 links the sensor signals based on maintenance history information. A phenomenon pattern classification criteria creation unit 107 creates classification criteria for classifying phenomenon patterns based on the extracted phenomenon patterns and work keywords included in the maintenance history information to which the sensor signals that are the basis of the phenomenon patterns are linked. A phenomenon pattern classification unit 108 classifies the phenomenon patterns based on the classification criteria. A diagnostic model creation unit 109 creates a diagnostic model for estimating work keywords to be presented to a maintenance worker based on the classified phenomenon patterns and work keywords" (see abstract). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2015-148867 A Summary of the Invention [Problem to be solved by the invention]

[0005] Fluid machinery used in chemical plants and power plants may not only operate continuously during plant operation, but may also change its operating state, such as from operation to stop and from stop to operation. When an existing anomaly cause estimation system is applied to such fluid machinery, the measurement values ​​from the fluid machinery used to estimate the cause of anomaly fluctuate significantly in the above-mentioned transient states, such as from operation to stop and from stop to operation. For this reason, there is a possibility that the fluid machinery may be estimated as abnormal when in fact there is no abnormality, so it is necessary to distinguish between a state in which an abnormality actually occurs and the above-mentioned transient state.

[0006] An object of the present invention is to provide a diagnostic device which distinguishes between an abnormality and a transient state and estimates the cause of the abnormality. [Means for solving the problem]

[0007] In order to solve the above problems, the diagnostic device of the present invention comprises a measurement value input unit that inputs each measurement value obtained from a plurality of sensors equipped in a target device, an abnormality determination unit that determines whether each measurement value is abnormal or not, and an abnormality cause estimation unit that estimates the cause of the abnormality of the target device based on the result of the abnormality determination unit's determination of whether each measurement value is abnormal or not, and the abnormality cause estimation unit estimates the cause of the abnormality of the target device using a combination of the presence or absence of abnormality of each measurement value obtained from the abnormality determination unit and an abnormality cause database in which combinations of measurement values ​​in which an abnormality has been detected and abnormality causes including the target device being in a transient state are stored in advance in correspondence with each other. Effect of the Invention

[0008] According to the present invention, it is possible to provide a diagnostic device that distinguishes between an actual abnormal state and a transient state and estimates the cause of the abnormality. [Brief description of the drawings]

[0009] [Figure 1] 1 is an overall configuration diagram of an embodiment of an abnormality cause estimation system according to the present invention; [Diagram 2] 1 is a diagram showing an overall structure of a centrifugal compressor as an example of a monitored device in the present invention. FIG. [Diagram 3] 1 is an example of an abnormality cause estimating device according to an embodiment of the present invention. [Figure 4] 4 is an example of time-series data of measurement values ​​according to an embodiment of the present invention. [Diagram 5] 4 is an example of values ​​recorded in a normal state measurement value database according to an embodiment of the present invention. [Figure 6] 13 is an example of a result of determining whether or not there is an abnormality in a time series of a transient state according to an embodiment of the present invention. [Figure 7A] 4 is an example of an initial state of data in an anomaly cause database in the embodiment of the present invention. [Figure 7B] 13 is an example of data in an anomaly cause database after updating in the embodiment of the present invention. [Figure 8] 13 is a diagram showing an example of a screen display of a diagnosis result output in the embodiment of the present invention. [Figure 9] 2 is a process flow according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, the embodiments of the present invention will be described in detail with reference to the drawings as appropriate. Note that, although the same reference numerals are used for equivalent components throughout the drawings, in some cases, different reference numerals are used for equivalent components for convenience of the drawings and description. EXAMPLES

[0011] 1 is an overall configuration diagram of the anomaly cause estimation system of this embodiment, which is composed of a customer facility 120 having devices and equipment that are the subject of anomaly cause estimation, a cloud 140, and a maintenance company facility 160 having equipment that performs the main processing of anomaly cause estimation. Note that the facility referred to here does not simply refer to a building, but is a general term that includes multiple structures having devices, equipment, or facilities and the areas in which they are installed. The customer facility 120 has a plant system 121 having monitored equipment 122 that is the target of anomaly cause estimation, an operation data collection device 123 that collects various operation data in the plant system 121, and a customer PC (Personal Computer) 124 that checks the operation data collected by the operation data collection device 123 and uploads the operation data to a cloud 140 (described later) via a public line or a dedicated line. In addition, by accessing the cloud 140, the customer PC 124 can view the results of anomaly cause estimation made by a maintenance company facility 160 (described later).

[0012] The cloud 140 includes a storage device 141 and a dashboard 142 , where the storage device 141 stores operation data from the customer facility 120 and the dashboard 142 displays the results of anomaly cause estimation performed by the maintenance company facility 160 . The maintenance company facility 160 has a dedicated hard disk drive (HDD) 161 for storing operation data and the like from the customer, and an analysis PC (Personal Computer) 162, which is an analysis device for estimating the cause of an abnormality using the operation data. Note that the above-mentioned customer PC and analysis PC are not limited to so-called personal computers, and may be any computer capable of executing the processing assumed in this embodiment, such as a general-purpose computer for business use or a laptop computer. 1 will roughly explain the flow of the process. Various operation data are collected as measurement data from monitored equipment 122 in plant system 121 at customer facility 120 by operation data collector 123, and are uploaded to storage device 141 in cloud 140 by customer PC 124. At maintenance company facility 160, the operation data in storage device 141 is downloaded to dedicated HDD 161, and anomaly cause estimation is performed at analysis PC 162. The results are not only displayed on analysis PC 162, but are also output to dashboard 142. At customer facility 120, the results of anomaly cause estimation output to dashboard 142 are viewed on customer PC 124.

[0013] In this embodiment, it is assumed that the monitored device 122 is a centrifugal compressor, which is a typical fluid machine, and FIG. 2 is a diagram showing an example of the overall structure of such a centrifugal compressor. In FIG. 2, the centrifugal compressor 201 includes a casing 202 formed into a cylindrical shape or the like and serving as a stationary part, and a rotor 205 is rotatably supported in the casing 202 by a journal bearing 203 and a thrust bearing 204. The rotor 205 is provided with a plurality of impellers 206, and the fluid is compressed by the rotation of the impellers 206. The centrifugal compressor 122 is provided with sensors (not shown) at various locations, and various operating data are acquired and transmitted to the operating data collection device 123. Examples of centrifugal compressors include those described in, for example, JP 2019-2361 A.

[0014] Next, the configuration of an abnormality cause estimation device 301 that estimates an abnormality and its cause from various operational data in the abnormality cause estimation system of this embodiment is shown in Fig. 3. The abnormality cause estimation device 301 is realized by the analytical PC 162 in Fig. 1, and is configured by application software, dedicated hardware, or a combination of these.

[0015] In FIG. 3, a measurement value input unit 302 receives measurement values ​​from various sensors installed in a monitored device 122 (corresponding to the centrifugal compressor 201 in FIG. 2 in this embodiment), and each measurement value is stored in a storage device 141 via an operation data collection device 123 and a customer PC 124, and then downloaded to a dedicated HDD 161 and input to an analysis PC 162. 4 shows an example of time series data 400 of input measured values, which includes the operating state (operating / stopped) 401 of the centrifugal compressor 201 at each time 410, and measured values ​​such as shaft vibration 402, suction temperature 403, discharge temperature 404, suction pressure 405, discharge pressure 406, and rotation speed 407 obtained from sensors (not shown) installed in the centrifugal compressor 201. Note that in an actual centrifugal compressor, etc., about 80 to 100 types of measured values, including those other than those mentioned above, are measured, but they are described in a simplified manner in this specification and the like.

[0016] Here, it can be seen that the time period 411 in which the operating state 401 switches from operating to stopped at time 410, or the time period 412 in which the operating state switches from stopped to operating, is a transient state, and each measurement value has a large fluctuation compared to the operating state up to that point.

[0017] A performance calculation unit 303 calculates polytropic efficiency as the performance of the fluid machinery from the measurement values ​​inputted by the measurement value input unit 302 . Anomaly determination unit 304 has a function of determining the presence or absence of an abnormality in the measurement value inputted by the measurement value input unit 302 and the polytropic efficiency (performance) calculated by the performance calculation unit 303. Here, an abnormality in the measurement value and performance of the sensor means that the measurement value and performance of the sensor deviate from a predetermined normal range, and if left as is, the normal function of the device may be impaired or may already be impaired.

[0018] The abnormality determination unit 304 acquires the measured values ​​and performance of the monitored device 122 (centrifugal compressor 201) under normal conditions from the normal condition measured value database 305, and determines whether or not there is an abnormality in the measured values ​​and the performance calculated from the measured values, based on the normal conditions. FIG. 5 shows examples of values ​​recorded in the normal condition measured value database 305, and abnormalities can be determined based on the magnitude of the difference between the actual measured values ​​and these normal conditions values ​​as the reference. Also, like the time series data of the measured values ​​in FIG. 4, multiple measured values ​​under normal conditions can be stored in the normal condition measured value database 305, and statistical processing (for example, Hotelling's T) can be performed using the population of these normal condition measured value data and the actual measured values. 2 The presence or absence of abnormalities may be determined by the following method.

[0019] Fig. 6 shows an example of the result of the abnormality determination unit 304 determining whether or not there is an abnormality in a time series of a transient state. Specifically, the abnormality determination is performed using data around the timing (411, 412, etc.) when the operation and stop are switched in the time series data of Fig. 4. In this example, it is shown that the values ​​of the shaft vibration 600, shaft movement 601, thrust bearing temperature 602, suction pressure 605, discharge pressure 606, rotation speed 607, and polytropic efficiency 608 are abnormal in the transient state, and the oil supply temperature 603 and the oil supply pressure 604 do not show abnormal values ​​in the transient state. In this example, it is shown that there are many measurement values ​​and performance values ​​of sensors that are determined to be abnormal in the transient state, and in some cases, the measurement values ​​of all sensors show abnormalities in the transient state.

[0020] The determination result by anomaly determination unit 304 is output to anomaly cause database 307 as additional data in a transient state during the initial preparation stage of this system (initial setup, etc.), and when this system actually performs anomaly cause estimation, this determination result is output to anomaly cause estimation unit 306.

[0021] The abnormality cause database 307 stores the relationship between the causes of abnormalities in the fluid machinery and the sensors and performance in which the abnormalities appear, and is, for example, a database created by learning based on past actual data in the plant. If it is a new plant, a database from another plant having a similar configuration may be used at the introduction stage and may be updated sequentially.

[0022] 7A shows an example of the initial state of a data table recorded in anomaly cause database 307. For example, if there is an abnormality in the sensor measurement value of axial vibration 700 and the polytropic efficiency value, then flow path dirt 721 is assumed as the abnormality cause 720. Similarly, if there are abnormalities in the measurement values ​​of axial movement 701, thrust bearing temperature 702, and discharge temperature 704 and the polytropic efficiency value, then internal leak 722 is assumed as the abnormality cause 720.

[0023] Here, in the initial state shown in Fig. 7A, the transient state of the target system is not taken into consideration, so in the initial stage of estimating the cause of anomaly, table data of measurement values ​​and performance values ​​that show abnormal values ​​in the transient state shown in Fig. 6 is added to the table in Fig. 7A to construct anomaly cause database 307 having table data shown in Fig. 7B. That is, Fig. 7B is a combination of table data 730 in Fig. 7A and table data 740 in Fig. 6.

[0024] The anomaly cause estimation unit 306 estimates the cause of the anomaly based on the anomaly determination result received from the anomaly determination unit 304 and the anomaly cause database 307. That is, the anomaly cause estimation unit 306 refers to the anomaly cause database 307, searches for a combination of the same pattern as the combination of the presence or absence of anomaly for each measurement value and performance value obtained from the anomaly determination unit 304, and when it detects a matching combination, it outputs the anomaly cause specified in the anomaly cause database 307 for that combination.

[0025] For example, if abnormalities are found in the values ​​of suction temperature 703, suction pressure 705, discharge pressure 706, rotation speed 707, and polytropic efficiency 708, it is estimated that the cause of the abnormality is a decrease in efficiency 731. Also, if abnormalities are found in the values ​​of shaft vibration 700, shaft movement 701, thrust bearing temperature 702, suction pressure 705, discharge pressure 706, rotation speed 707, and polytropic efficiency 708, it can be determined that this is due to a transient state. This makes it possible to distinguish between a true abnormality and an abnormality in the measured values ​​in a transient state, and to estimate the cause of the abnormality.

[0026] As described above, in an actual system, there may be as many as 80 to 100 types of measurement value items (for example, items from axial vibration 700 to polytropic efficiency 708 in the table data of Fig. 7B), and there may be cases where the actual measurement value item showing an abnormal value does not completely match the combination of abnormal items in the table data of Fig. 7B. In such cases, multiple combinations of similar items may be extracted from the table data of Fig. 7B, and the similarity or match rate with the combination of abnormal actual measurement values ​​may be calculated using a known method, and the combinations may be presented in order of similarity (match rate).

[0027] The result output unit 308 receives the measurement values ​​and performance anomaly judgment results and the estimated anomaly causes from the anomaly cause estimation unit 306, and outputs them as diagnostic results using graphs, tables, etc. so that they are visually easy to understand. These diagnostic results can be displayed on a display device (such as the screen of the analytical PC 162), or can be output to the dashboard 142 of the cloud 140 and viewed on the customer PC 124.

[0028] FIG. 8 shows an example of the screen display of the above-mentioned diagnostic result output on the analysis PC 162 or the dashboard 142. The target period 800 is the period during which the diagnosis is performed, that is, the period during which the measurement data is acquired. Here, it is one month from March 1st to March 31st. 810, 820, and 830 indicate events in which abnormalities were found in the measurement data during the target period. For example, event 1 (810) indicates that abnormalities were detected in the measured values ​​of shaft vibration, rotation speed, etc. from 14:11 to 14:13 on March 2nd, and the results of cause estimation indicate that the probability of flow path fouling, efficiency reduction, and poor lubrication is 88%, 65%, and 43%, respectively. In order to improve visibility, these probabilities may be shown in a bar graph.

[0029] In addition, in Event 3 (830), abnormalities were detected in the measurements of fuel supply pressure, discharge pressure, etc. from 16:41 to 16:43 on March 25th, but the cause was estimated to be due to a transient state of the target equipment, and not an abnormality in the equipment itself.

[0030] 9 shows a process flow in the anomaly cause estimation system of this embodiment. First, data in a transient state is extracted from the measurement data acquired from the target device (step S911). Then, an anomaly is judged using the measurement data in the transient state (step S912), and a combination of sensors (measurement values) that showed an abnormal value in the transient state (corresponding to FIG. 6) is acquired (step S913), and is added to the initial state (corresponding to FIG. 7A) of the anomaly cause database 307, thereby constructing an anomaly cause database 307 (corresponding to FIG. 7B) including the transient state. Steps S911 to S914 (pre-processing 910) up to this point are performed when the present system is introduced, etc., and do not necessarily have to be performed every time, but may be performed periodically as necessary.

[0031] After the above pre-processing 910 is performed, the process proceeds to the actual main process 920 (steps S921 to S923) of estimating the cause of anomaly. First, an anomaly judgment is performed for all measurement data including the transient state using the normal state measurement database 305 (FIG. 5) (step S921), and for the combination of measurement values ​​(and performance values) judged to be abnormal, an anomaly cause is estimated using the anomaly cause database 307 (FIG. 7B) (step S922). Then, the estimation result is output in the form of a screen display such as that shown in FIG. 8 (step S923). In this way, it is possible to perform an anomaly judgment and estimate the cause of the anomaly while taking into account the transient state of the target device.

[0032] The present invention is not limited to the above-mentioned embodiment, and includes various modified examples. For example, the storage device 141 and the analysis PC 162 provided in the cloud 140 and the maintenance company facility 160 in Fig. 1 may be provided in the customer facility 120, so that the customer himself can estimate the cause of the abnormality. Furthermore, the operating data may be sent by email from the customer to the maintenance company, and the result of the estimation of the cause of the abnormality may be provided to the customer as a document such as a report.

[0033] The above-mentioned embodiment has been described in detail to clearly explain the present invention, and is not necessarily limited to those including all of the described configurations. The above-mentioned configurations, functions, etc. may be realized by designing a part or all of them as an integrated circuit, for example. The above-mentioned configurations, functions, etc. may be realized by software, in which a processor interprets and executes a program that realizes each function. [Explanation of symbols]

[0034] 100: Anomaly Cause Identification System 200: Centrifugal compressor 301: Abnormality cause estimation device 302: Measurement value input section 303: Performance calculation section 304: Abnormality determination section 305: Normal measurement value database 306: Abnormal cause estimation section 307: Anomaly Cause Database 308: Result output section

Claims

1. A diagnostic device that estimates a cause of an abnormality based on a measurement value of a target device, a measurement value input unit for inputting the measurement values ​​obtained from a plurality of sensors provided in the target device; an abnormality determination unit that determines whether or not the measurement value is abnormal; an anomaly cause estimation unit that estimates a cause of the anomaly of the target device based on a result of the anomaly determination unit as to whether or not the measurement value is abnormal; having the anomaly cause estimation unit estimates a cause of the anomaly of the target device by using an anomaly cause database in which a combination of the measurement values ​​indicating the presence or absence of an anomaly acquired from the anomaly determination unit, a combination of the measurement values ​​in which an anomaly has been detected, and causes of the anomaly, including the target device being in a transient state, are stored in advance in association with each other; A diagnostic device characterized by:

2. The diagnostic device according to claim 1, a normal state measurement database in which representative values ​​of the measurement values ​​of the target device under normal conditions are recorded; the abnormality determination unit compares a statistic calculated from the measurement value from the measurement value input unit with a statistic calculated from a representative value of the normal state measurement database, and performs abnormality determination based on a difference between the two. A diagnostic device characterized by:

3. The diagnostic device according to claim 1, a normal state measurement database in which a plurality of time-series values ​​of the measurement values ​​of the target device under normal conditions are recorded; the abnormality determination unit performs abnormality determination by statistical processing using the measurement value from the measurement value input unit and the plurality of values ​​in the normal state measurement database. A diagnostic device characterized by:

4. The diagnostic device according to claim 2 or 3, a performance calculation unit that calculates a performance value of the target device using the measurement value obtained from the sensor; The normal state measurement database records performance values ​​of the target device under normal conditions, the abnormality determination unit performs abnormality determination with respect to the measurement value from the measurement value input unit and the performance value from the performance calculation unit by referring to the normal state measurement database; A diagnostic device characterized by:

5. The diagnostic device according to claim 4, the anomaly cause database stores in advance a combination of the measurement value and the performance value in which an anomaly has been detected and an anomaly cause including the target device being in a transient state, in association with each other; the anomaly cause estimation unit estimates a cause of the anomaly of the target device by using a combination of the presence or absence of an anomaly of the measurement value and the performance value acquired from the anomaly determination unit and the anomaly cause database; A diagnostic device characterized by:

6. The diagnostic device according to claim 5, the anomaly cause estimation unit refers to the anomaly cause database, and searches for a combination of the same pattern as a combination of the presence or absence of an anomaly of the measurement value and the performance value acquired from the anomaly determination unit; When a matching combination is detected, the cause of the abnormality for that combination, which is specified in the abnormality cause database, is output; if no matching combination is found, outputting anomaly causes defined in said anomaly cause database for similar combinations in order of similarity; A diagnostic device comprising:

7. 7. The diagnostic device according to claim 6, a result output unit that displays the abnormality cause output from the abnormality cause estimation unit on a display device using a table or graph; A diagnostic device comprising:

8. 7. The diagnostic device according to claim 6, The target equipment is a fluid machine; A diagnostic device comprising:

9. A diagnostic method for estimating a cause of an abnormality based on a measurement value of a target device, comprising: inputting the measured values ​​obtained from a plurality of sensors provided in the target device; Determining whether the measurement value is abnormal or not; a combination of the measurement values ​​in which an abnormality is detected and a cause of the abnormality including the target device being in a transient state are stored in advance in correspondence with each other, and the cause of the abnormality of the target device is estimated by comparing and referring to the combination of the measurement values ​​determined to have an abnormality or not; A diagnostic method characterized by:

10. A diagnostic system having the diagnostic device according to claim 4, The diagnostic system includes a communication means at a customer facility that can access a cloud and a diagnostic device at a maintenance company facility, The communication means of the customer facility collects the measurement values ​​used for maintenance from the target devices in the plant system and uploads them to a storage device of the cloud; The diagnostic device at the maintenance company facility is the diagnostic device according to claim 4, which downloads the measurement values ​​uploaded to the storage device and outputs the presence or absence of an abnormality in the target device and the cause of the abnormality as a diagnostic result to a dashboard of the cloud. A diagnostic system comprising:

Images