Abnormality determination device, abnormality determination method, abnormality determination system, control method for abnormality determination system, and program

Abstract

To properly determine an abnormality on the basis of time-series data of status parameters during operation.SOLUTION: An abnormality determination device for determining an abnormality of a device in which a temporal change of status parameters during operation indicates a specific profile waveform comprises a time-series data acquisition unit, a slide window setting unit, and an abnormality determination unit. The time-series acquisition unit acquires time-series data including a plurality of temporarily continuous data about the status parameters. The slide window setting unit sets a length of the slide window indicating the number of data included in the time-series data. The abnormality determination unit extracts a plurality of reference time-series data corresponding to the time-series data from reference data indicating a temporal change of the status parameters corresponding to a normal operation status of a device, and determines the presence and absence of an abnormality in the device by comparing the time-series data having the length of the slide window set by the slide window setting unit with the plurality of reference time-series data while sliding the slide window.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present disclosure relates to an apparatus for determining an abnormality and a method for determining an abnormality. [Background technology]

[0002] For example, equipment used in plant facilities may experience abnormalities such as breakdowns or malfunctions. While equipment abnormalities can be detected through inspections performed during maintenance, such inspections require the equipment to be shut down while in operation, disrupting the equipment's operation schedule. Therefore, it is desirable to detect the status of equipment during operation using sensors or the like and to determine equipment abnormalities in advance or at an early stage based on the detection results.

[0003] A technology relating to such abnormality detection of equipment during operation is, for example, Patent Document 1. In this document, time-series data is acquired from multiple sensors attached to target equipment that continuously performs the same operation, such as a vibrating conveyor, and the data is combined to obtain synthetic data, which is then compared with learned synthetic data from normal times to determine abnormalities in the target equipment. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2021-38946 Summary of the Invention [Problem to be solved by the invention]

[0005] In the above Patent Document 1, when making an abnormality judgment, for example, a sliding window is used to decompose the composite data related to the time series data and the composite data in normal times into partial data, and then judgment is made. However, if the sliding window is not set appropriately, there is a problem that the accuracy in making an abnormality judgment decreases.

[0006] At least one embodiment of the present disclosure has been made in consideration of the above-mentioned circumstances, and aims to provide an abnormality determination device, an abnormality determination method, an abnormality determination system, a control method for an abnormality determination system, and a program that can appropriately determine an abnormality based on time-series data of state parameters during operation. [Means for solving the problem]

[0007] In order to solve the above problem, an abnormality determination device according to at least one embodiment of the present disclosure includes: An abnormality determination device for a device in which a time change in a state parameter during operation shows a specific profile waveform, a time-series data acquisition unit for acquiring time-series data including a plurality of pieces of time-sequential data for the state parameter; a sliding window setting unit that sets a length of a sliding window that indicates the number of data included in the time series data; an abnormality determination unit for extracting a plurality of reference time series data corresponding to the time series data from reference data indicating time changes of the state parameter corresponding to a normal operating state of the device, and for determining whether or not there is an abnormality in the device by comparing the time series data having the length of the sliding window set by the sliding window setting unit with the plurality of reference time series data while sliding the sliding window; Equipped with.

[0008] In order to solve the above problem, an abnormality determination method according to at least one embodiment of the present disclosure includes: A method for determining an abnormality in a device in which a time change in a state parameter during operation shows a specific profile waveform, comprising: an acquiring step of acquiring time-series data including a plurality of time-sequential data for the state parameter; a sliding window setting step of setting a length of a sliding window indicating the number of data included in the time series data; an abnormality determination step of extracting a plurality of reference time series data corresponding to the time series data from reference data indicating time changes of the state parameters corresponding to a normal operating state of the device, and determining whether or not there is an abnormality in the device by comparing the time series data having the length of the sliding window set in the sliding window setting step with the plurality of reference time series data while sliding the sliding window; Equipped with.

[0009] In order to solve the above problem, a program according to at least one embodiment of the present disclosure includes: Computer, an acquisition means for acquiring time-series data including a plurality of time-sequential data for a state parameter of a device whose time change in the state parameter exhibits a specific profile waveform during operation; a sliding window setting means for setting a length of a sliding window indicating the number of data included in the time series data; an abnormality determination means for extracting a plurality of reference time series data corresponding to the time series data from reference data indicating time changes of the state parameters corresponding to a normal operating state of the device, and for determining whether or not there is an abnormality in the device by comparing the time series data having the length of the sliding window set by the sliding window setting means with the plurality of reference time series data while sliding the sliding window; and make it work.

[0010] In order to solve the above problem, an abnormality determination system according to at least one embodiment of the present disclosure includes: An abnormality determination system including an abnormality determination device for a device whose time-varying change in a state parameter during operation shows a specific profile waveform, and a client terminal, The abnormality determination device a time-series data acquisition unit for acquiring time-series data including a plurality of pieces of time-sequential data for the state parameter; a sliding window setting unit that sets a length of a sliding window that indicates the number of data included in the time series data; an abnormality determination unit that, in response to a request from the client terminal, extracts a plurality of reference time series data corresponding to the time series data from reference data that indicates time changes of the state parameters corresponding to a normal operating state of the device, and determines whether or not an abnormality exists in the device by comparing the time series data having the length of the sliding window set by the sliding window setting unit with the plurality of reference time series data while sliding the sliding window; Equipped with.

[0011] In order to solve the above problem, a control method for an abnormality determination system according to at least one embodiment of the present disclosure includes: A control method for an abnormality determination system including an abnormality determination device for a device whose time-varying state parameter during operation shows a specific profile waveform and a client terminal, comprising: The abnormality determination device a time-series data acquisition step for acquiring time-series data including a plurality of time-sequential data for the state parameter; a sliding window setting step of setting a length of a sliding window indicating the number of data included in the time series data; an abnormality determination step of extracting, in response to a request from the client terminal, a plurality of reference time series data corresponding to the time series data from reference data indicating time changes of the state parameters corresponding to a normal operating state of the device, and comparing the time series data having the length of the sliding window set in the sliding window setting step with the plurality of reference time series data while sliding the sliding window, thereby determining whether or not there is an abnormality in the device; Execute. [Effects of the Invention]

[0012] According to at least one embodiment of the present disclosure, it is possible to provide an abnormality determination device, an abnormality determination method, an abnormality determination system, a control method for an abnormality determination system, and a program that can appropriately determine an abnormality based on time-series data of state parameters during operation. [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a schematic diagram illustrating an abnormality determination device according to an embodiment, together with a device to be determined; [Figure 2] 2 is a diagram showing an example of time variations in state parameters of a normal device detected by the sensor in FIG. 1. FIG. [Figure 3] 2 is a block diagram showing a hardware configuration of the abnormality determination device of FIG. 1. FIG. [Figure 4] 2 is a block diagram showing a functional configuration of the abnormality determination device of FIG. 1. FIG. [Figure 5A] 5 is a flowchart showing a method for extracting reference time series data by the reference time series data acquisition unit 162 of FIG. 4. [Figure 5B] 5 is a flowchart showing an abnormality determination method in the abnormality determination unit of FIG. 4. [Figure 6A] FIG. 10 is a diagram schematically illustrating a plurality of reference time series data extracted from reference data. [Figure 6B] FIG. 10 is a diagram illustrating a process of acquiring time-series data by applying a sliding window. [Figure 7] FIG. 6 is a diagram schematically showing positions in the virtual space identified in step S202 of FIG. 5. [Figure 8] FIG. 10 is a schematic diagram showing the position of a sliding window in reference data corresponding to model data. [Figure 9] FIG. 10 is a diagram schematically illustrating positions corresponding to example data in a virtual space. [Figure 10] 4 is a graph showing an example of abnormality determination performed by the abnormality determination device according to an embodiment. [Figure 11] FIG. 2 is a diagram schematically illustrating a data acquisition period with respect to a change over time in a state parameter of a device. [Figure 12] 10 is a diagram schematically illustrating an appliance operation period from which a plurality of reference time-series data are extracted from the reference data. FIG. [Figure 13] FIG. 10 is a diagram showing changes over time in state parameters in each operation mode. [Figure 14]FIG. 10 is a block diagram showing a functional configuration of an abnormality determination device 100 according to another embodiment. [Figure 15] 1 is a configuration diagram of an abnormality determination system according to an embodiment; [Figure 16] FIG. 16 is a block diagram of the abnormality determination system of FIG. DETAILED DESCRIPTION OF THE INVENTION

[0014] Hereinafter, several embodiments of the present disclosure will be described 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 present disclosure.

[0015] FIG. 1 is a schematic diagram showing an anomaly determination device 100 according to one embodiment, together with a device 1 that is the target of determination. The device 1 is any device in which the time change of a state parameter P during operation shows a specific profile waveform PF. The device 1 is provided with a sensor 2 for detecting the state parameter P that indicates the state of the device 1. The sensor 2 continuously detects the state parameter P at a predetermined sampling frequency, thereby making it possible to acquire the time change of the state parameter P. When the device 1 operates at a predetermined timing, the state parameter P detected by the sensor 2 changes.

[0016] In this embodiment, as an example of the device 1 to be determined, a control valve that periodically operates to adjust the amount of fuel (pulverized coal) supplied to a gasifier that generates gas from coal to drive a gas turbine in an integrated coal gasification combined cycle (IGCC) power plant will be described. Such a control valve is a valve (pulverized coal supply hopper pressure reducing / exhausting valve) that reduces the pressure and exhausts the pressure of a pulverized coal supply hopper that supplies pulverized coal as fuel to the gasifier. Such a control valve is periodically operated at a preset timing.

[0017] FIG. 2 is a diagram showing an example of a change over time in the state parameter P of a normal device 1 detected by the sensor 2 of FIG. 1. This example shows the change over time in the state parameter P when the device 1, which is a control valve, periodically opens at a preset timing, as described above. FIG. 2 shows the change over time in the state parameter P when the device 1 is normal (when no abnormality is present), and shows how multiple profile waveforms PF corresponding to the periodic operation timings of the device 1 appear at approximately equal intervals. Each profile waveform PF includes a rising portion PF1 in which the state parameter P increases from a base value P0 in response to the opening operation of the control valve, a peak portion PF2 corresponding to the fully open state of the control valve, and a falling portion PF3 in which the state parameter P decreases toward the base value P0 in response to the closing operation of the control valve, and has approximately the same shape as each other.

[0018] Next, we will explain the specific configuration of the abnormality determination device 100. Fig. 3 is a block diagram showing the hardware configuration of the abnormality determination device 100 of Fig. 1.

[0019] The abnormality determination device 100 is configured as an arithmetic processing device such as a computer. The hardware configuration of the abnormality determination device 100 includes an input unit 110, a storage unit 120, a calculation unit 130, and an output unit 140, as shown in FIG.

[0020] The input unit 110 is configured to input various pieces of information necessary for the arithmetic processing performed in the abnormality determination device 100. In this embodiment, the input unit 110 is configured as an interface device capable of receiving data transmitted from the sensor 2 provided in the device 1, and may additionally include a human interface such as a mouse, keyboard, and touch panel that can be operated by an operator, and interface devices for receiving data from other devices.

[0021] The storage unit 120 is configured to store various types of information necessary for the arithmetic processing performed in the abnormality determination device 100. The storage unit 120 is configured from a computer-readable storage medium including at least one of a RAM (Random Access Memory) and a ROM (Read Only Memory). The various types of information stored in the storage unit 120 include programs that enable these hardware configurations to function as the abnormality determination device 100.

[0022] The calculation unit 130 is configured to perform various calculations of the abnormality determination device 100, and is configured to include, for example, a CPU (Central Processing Unit). The calculation unit 130 reads out the programs stored in the storage unit 120 into a RAM or the like, and executes information processing and calculation processing, thereby realizing various functions of the abnormality determination device 100.

[0023] The program executed by the calculation unit 130 may be stored in the storage unit 120 as described above, or may be pre-installed in a ROM or other storage medium, provided in a state stored in a computer-readable storage medium, or distributed via wired or wireless communication means, etc. Examples of computer-readable storage media include magnetic disks, magneto-optical disks, CD-ROMs, DVD-ROMs, and semiconductor memories.

[0024] The output unit 140 is configured to perform output based on the calculation result in the calculation unit 130. In this embodiment, the output unit 140 outputs a determination result regarding the presence or absence of an abnormality in the device 1 as the output of the abnormality determination device 100. This determination result may be recognized by an operator in a form that can be recognized by humans, such as a display or an alarm, or may be input as a signal corresponding to the determination result to another device and used for various controls.

[0025] Fig. 4 is a block diagram showing the functional configuration of the abnormality determination device 100 of Fig. 1. The abnormality determination device 100 includes a data monitoring unit 150, a reference data storage unit 160, and an abnormality determination unit 170. The block diagram shown in FIG. 4 is an example showing the functional configuration of the abnormality determination device 100 in accordance with the following description, and the blocks may be integrated with each other or further subdivided.

[0026] The data monitoring unit 150 includes a time-series data acquiring unit 152 for acquiring time-series data Dt including a plurality of pieces of time-sequential data for the state parameter P. In this embodiment, as described above with reference to FIG. 1, the device 1 is provided with a sensor 2 for detecting the state parameter P, and the time-series data acquiring unit 152 acquires the time-series data Dt by continuously receiving the detection results of the sensor 2 detected at a predetermined sampling frequency.

[0027] The reference data storage unit 160 is configured to store reference data Dref indicating a time change of the state parameter P corresponding to the normal operating state of the device 1, and includes various computer-readable storage media such as hardware and memory. The reference data Dref is acquired in advance using the device 1 in a normal operating state, and is data indicating a time change of the state parameter P including a plurality of profile waveforms PF of approximately the same shape corresponding to the operation of the device 1, as exemplified in FIG. 2.

[0028] The reference data storage unit 160 includes a reference time series data acquisition unit 162. The reference time series data acquisition unit 162 is configured to acquire, from the reference data Dref stored in the reference data storage unit 160, reference time series data that serves as a reference when comparing with the time series data acquired by the time series data acquisition unit 152 included in the data monitoring unit 150.

[0029] Each data included in the reference data Dref is acquired under the same conditions as the time series data Dt acquired by the time series data acquiring unit 152. For example, the sampling frequency when acquiring each data included in the reference data Dref is set to be the same as the sampling frequency of the time series data Dt acquired by the time series data acquiring unit 152 (the sampling frequency of the sensor 2).

[0030] The abnormality determination unit 170 determines whether or not an abnormality exists in the device 1 by comparing the time series data acquired by the time series data acquisition unit 152 of the data monitoring unit 150 with the reference time series data acquired by the reference time series data acquisition unit 162 from the reference data Dref stored in the reference data storage unit 160. When there is no abnormality in the device 1, the time series data acquired by the time series data acquisition unit 152 of the data monitoring unit 150 have approximately the same shape as each profile waveform PF, as shown in FIG. 2 . However, when there is an abnormality in the device 1, at least one profile waveform PF included in the time series data acquired by the time series data acquisition unit 152 of the data monitoring unit 150 will be deformed to a considerable extent compared to a normal case. The abnormality determination unit 170 compares the time series data acquired by the time series data acquisition unit 152 of the data monitoring unit 150 with the reference time series data acquired by the reference time series data acquisition unit 162 of the reference data storage unit 160, and thereby evaluates the shape of the profile waveform PF included in the time series data to determine whether or not an abnormality exists.

[0031] The sliding window setting unit 180 sets information related to the length of the sliding window to be applied to the reference data Dref and the data received from the sensor 2, the width by which the sliding window is slid, etc. As an example of a method for setting the information related to the sliding window, the setting may be made based on information input by the user via the input unit 110, or the setting may be made based on the profile waveform PF to be applied.

[0032] Next, a detailed description will be given of an abnormality determination method performed by the abnormality determination device 100 having the above configuration. Fig. 5A is a flowchart showing a method for extracting reference time series data by the reference time series data acquisition unit 162 in Fig. 4, and Fig. 5B is a flowchart showing an abnormality determination method in the abnormality determination unit 170 in Fig. 4. 5B, steps S202 and thereafter may be performed by the abnormality determination unit 170, and steps S200 and S201 may be performed by the data monitoring unit 150.

[0033] First, as shown in Fig. 5A, the reference time series data acquisition unit 162 acquires the reference data Dref stored in the reference data storage unit 160 (step S100). Then, as will be described later with reference to Fig. 6A, the sliding window set by the sliding window setting unit 180 is applied to the reference data Dref acquired in step S100 (step S101), and reference time series data is extracted (step S102). The reference time series data extracted in step S102 is stored in the reference data storage unit 160 so as to be readable as appropriate (step S103).

[0034] Here, FIG. 6A shows a plurality of reference time series data x extracted from the reference data Dref. (1) , x (2) , . . . In this example, multiple reference time series data x (1) , x (2) , ... is extracted so as to include ω pieces of time-sequential data corresponding to the length of the sliding window ω. Specifically, the reference time series data x (1) is the number of consecutive ω data ε in the reference data Dref (1) , ε (1) , ···, ε (ω) The reference time series data x is extracted to include (2) is the number of consecutive ω data ε in the reference data Dref (2) , ε (3) , ···, ε (ω+1) Specifically, as illustrated in FIG. 6A, for example, the length of the sliding window is ω=3, and the reference data is ξ (1) , ξ (2) , ···ξ (12) If the reference time series data is X (1) , X (2) , X (10) So the number is 10.

[0035] In this embodiment, a plurality of reference time series data x (1) , x (2), ... exemplify a case where a data group including ω pieces of temporally consecutive data is extracted by sliding the data by one piece at a time, but the sliding width may be set to two pieces of data or more. In this embodiment, adjacent reference time series data are extracted so that they overlap to some extent, but adjacent reference time series data may also be extracted so that they do not overlap.

[0036] 5B, the abnormality determination unit 170 first acquires the time series data Dt received from the sensor 2 provided in the device 1 from the time series data acquisition unit 152 of the data monitoring unit 150 (step S200). In step S200, the same sliding window used in step S101 of FIG. 5A is applied to acquire the time series data Dt (step S201).

[0037] 6B is a diagram schematically illustrating how time series data Dt is acquired by applying a sliding window. The time series data acquisition unit 152 sequentially acquires data transmitted from the sensor 2, and acquires time series data Dt such that the number of pieces of data thus acquired is equal to the length of the sliding window set by the sliding window setting unit 180 used to extract the reference time series data in step 101 of FIG. 5A. That is, a sliding window of the same length as the reference time series data is applied to the time series data Dt.

[0038] The length ω of the sliding window may be set based on the profile waveform PF. In this embodiment, the length ω of the sliding window may be set to have a time width greater than at least one of the rising portion PF1 and the falling portion PF3 of the profile waveform PF. If an abnormality occurring in the device 1 has a characteristic that is likely to be reflected in the rising portion PF1 and the falling portion PF3 of the profile waveform PF, setting the length ω of the sliding window in this manner makes it possible to suitably determine an abnormality in the device 1. The length ω of the sliding window may also be set to have a time width greater than that of the profile waveform PF. In this case, setting the length ω of the sliding window to have a time width greater than that of the profile waveform makes it possible to more suitably determine an abnormality in the device.

[0039] Next, the abnormality determination unit 170 calculates the position p in the virtual space V corresponding to the time series data Dt acquired in step S200 and the plurality of reference time series data x stored in the reference data storage unit 160 in step S103. (1) , x (2) , . . . in the virtual space V are identified (step S202).

[0040] 7 is a diagram schematically illustrating each position in the virtual space V identified in step S202 of FIG. 5B. The virtual space V is expressed as an ω-dimensional space corresponding to the length of the sliding window, but in FIG. 7, it is simply shown on a two-dimensional plane for ease of understanding.

[0041] In step S202, the time series data Dt acquired in step S201 is expressed in the virtual space V as a position p corresponding to an ω-dimensional vector whose components are each data included in the time series data Dt (see the black dots in FIG. 7). Also in step S202, the plurality of reference time series data x extracted in step S103 are expressed as a position p corresponding to an ω-dimensional vector whose components are each data included in the time series data Dt (see the black dots in FIG. 7). (1) , x (2) ,... are the reference time series data x in the virtual space V. (1) , x (2) , ... (1) is a set of multiple time-sequential data ε (1) , ε (1) , ···, ε (n) corresponds to an ω-dimensional vector with components, and the reference time series data x (2) is a set of multiple time-sequential data ε (2) , ε (3) , ···, ε (n+1) corresponds to an ω-dimensional vector with components TIFF0007792810000001.tif22170

[0042] The abnormality determination unit 170 uses the plurality of reference time series data x(1) , x (2) , ··· may be further aggregated and treated as a set matrix D expressed by the following equation: TIFF0007792810000002.tif10170

[0043] Next, the abnormality determination unit 170 calculates the degree of deviation L between the position p identified in step S202 and each of the positions p1, p2, ... (step S203). The degree of deviation means the distance between two points in the virtual space V, and in step S203, the degree of deviation L between the position p identified in step S202 and any one selected from the positions p1, p2, ... is calculated.

[0044] In one embodiment, in step S203, the abnormality determination unit 170 calculates the degree of deviation L between the position p identified in step S202 and the kth closest position from the position p among the positions p1, p2, ..., where k is an arbitrary natural number. For example, when k=1, the degree of deviation L between the position p identified in step S4 and the closest position from the position p (i.e., the nearest position) among the positions p1, p2, ..., identified in step S202 is calculated. In this embodiment, the smaller k is, the higher the sensitivity of abnormality determination is expected to be, but the value of k can be adjusted arbitrarily.

[0045] In another embodiment, in step S203, the abnormality determination unit 170 may calculate the degree of deviation L between the position p identified in step S202 and a representative position obtained by calculation from each of the positions p1, p2, .... In this case, the representative position may be obtained by statistical calculation of, for example, the average value of each of the positions p1, p2, ... identified in step S202.

[0046] In another embodiment, in step S203, the abnormality determination unit 170 may calculate the degree of deviation L between the position p identified in step S202 and the position pm corresponding to the model data Dm selected from the positions p1, p2, .... Here, the model data Dm is a set of the plurality of reference time series data x extracted in step S103. (1) , x(2) , ..., is selected as data that has a high degree of contribution to abnormality determination.

[0047] For example, multiple reference time series data x extracted from the reference data Dref (1) , x (2) , ... can be roughly divided into those that include at least a portion of the profile waveform PF among the reference data Dref, and those that do not include the profile waveform PF. The former include at least a portion of the profile waveform PF, and therefore contribute greatly to abnormality determination when compared with the time-series data Dt, whereas the latter do not include the profile waveform PF, and therefore contribute little to abnormality determination when compared with the time-series data Dt. In this case, the former is selected as the model data Dm.

[0048] 8 is a schematic diagram showing the position of the sliding window corresponding to the model data Dm in the reference data Dref. In FIG. 8, since at least a part of each of the sliding windows ω1 to ω3 includes the profile waveform PF, the reference time series data x corresponding to the sliding windows ω1 to ω3 (1) , x (2) , ... are treated as model data Dm. Specifically, the reference time series data corresponding to the sliding window ω1 mainly contains the rising portion PF1 of the profile waveform PF, and therefore contributes greatly to abnormality detection and is included in the model data Dm. The reference time series data corresponding to the sliding window ω2 mainly contains the peak portion PF2 of the profile waveform PF, and therefore contributes greatly to abnormality detection and is included in the model data Dm. Furthermore, the reference time series data corresponding to the sliding window ω3 mainly contains the falling portion PF3 of the profile waveform PF, and therefore contributes greatly to abnormality detection and is included in the model data Dm. On the other hand, the reference time series data corresponding to the sliding window ω4 does not contain the profile waveform PF, and therefore contributes little to abnormality detection and is not included in the model data Dm.

[0049] 9 is a diagram showing a schematic diagram of a position pm corresponding to model data Dm in a virtual space V. In FIG. 9, a plurality of reference time series data x (1) , x (2), ... are shown, and among them, those included in the model data Dm form cluster C. This is because, in the device 1 that performs periodic operations, as described above, profile waveforms PF having approximately the same shape appear periodically and repeatedly in the time-series data Dt when the device 1 is operating normally. In this embodiment, by calculating the degree of deviation from the positions pm included in the model data Dm in this way, it is possible to perform highly reliable abnormality determination.

[0050] Next, it is determined whether the deviation L calculated in step S203 is equal to or greater than a threshold value (step S204). The threshold value is a determination reference value that is set in advance for the deviation L to determine whether or not an abnormality exists. As a result, if the deviation L is equal to or greater than the threshold value (step S204: YES), the abnormality determination unit 170 determines that an abnormality exists in the device 1 (step S205). On the other hand, if the deviation L is less than the threshold value (step S204: NO), the abnormality determination unit 170 determines that no abnormality exists in the device 1 (step S206).

[0051] Fig. 10 is a graph showing an example of abnormality determination by the abnormality determination device 100 according to one embodiment. In this example, a control valve used in a boiler for generating steam to drive a generator in a power plant facility is used as the device 1 to be subjected to abnormality determination. As shown in Fig. 10, this power plant facility is operated so that the generator output is constant, and the control valve, which is the device 1, operates periodically. Accordingly, the aperture of the control valve, which is a state parameter of the device 1, is periodically opened and closed, resulting in a specific profile waveform PF.

[0052] 10 shows the generator output and the opening degree of the control valve (state parameter P) as well as the transition of the deviation calculated by the abnormality determination unit 170. The deviation changes over time, and an abnormality is determined when it exceeds a preset threshold value near times t1 and t2. In this example, in an actual power plant facility, an abnormality in the control valve becomes apparent when the behavior of the state parameter P changes near time t3. However, in FIG. 10, an abnormality can be determined at the precursory stage at time t1, before time t3, even though there is no change in the behavior of the state parameter P. This shows that the abnormality determination device 100 having the above configuration can determine an abnormality in the device 1 early and accurately.

[0053] In another embodiment, in step S200, the time-series data acquisition unit 152 may acquire the time-series data Dt during a period (data acquisition period T1) corresponding to the operation timing of the device 1. Here, Fig. 11 is a diagram schematically showing the data acquisition period T1 with respect to the change over time in the state parameters of the device 1. Like Fig. 2, Fig. 11 shows the change over time in the state parameters of the device 1, and shows multiple profile waveforms PF corresponding to the operation timing of the device 1. The data acquisition period T1 is set to correspond to each profile waveform PF, and each of the data acquisition periods T1 includes a profile waveform PF.

[0054] Such a data acquisition period T1 does not include a non-data acquisition period T2 in which the state parameter P indicating that the device 1 is not operating is approximately zero. In the non-data acquisition period T2, the state parameter P is approximately zero, so even if an abnormality occurs in the device 1, the impact is small. Therefore, by acquiring time-series data in the data acquisition period T1, the time-series data acquisition unit 152 can reduce the range in which the state parameter P included in the time-series data is approximately zero. Because such a range in which the state parameter P is approximately zero has only a minor impact on the abnormality determination result, excluding it from the time-series data can effectively reduce the calculation load on the abnormality determination unit 170 without reducing the accuracy of the abnormality determination.

[0055] 11, each profile waveform PF is accompanied by a small sub-waveform SFsub. The sub-waveform SFsub has a smaller peak size than the main profile waveform PF, and therefore has less impact on abnormality determination. Therefore, by setting the data acquisition period T1 to include the profile waveform PF but not the sub-waveform SFsub, the calculation load on the abnormality determination unit 170 can be effectively reduced without reducing the accuracy of the abnormality determination.

[0056] In another embodiment, the abnormality determination unit 170 may compare the time series data Dt with a plurality of reference time series data x (1) , x (2) , . . . are extracted from the reference data Dref, the reference data Dt is extracted from the device operation period including the profile waveform PF, and multiple reference time series data x (1) , x (2) , . . . may be extracted. Here, FIG. 12 shows a plurality of reference time series data x (1) , x (2) 1 is a diagram schematically illustrating an appliance operation period T3 from which the reference data Dref are extracted. The appliance operation period T3 is set as a period including a profile waveform PF that appears in response to the operation of the appliance 1 in the reference data Dref, and each of the profile waveforms includes the profile waveform PF.

[0057] Such an appliance operation period T3 does not include a non-appliance operation period T4 in which the state parameter indicating that the appliance 1 is not operating is approximately zero. In the appliance non-operation period T4, the state parameter P is approximately zero, so even if an abnormality occurs in the appliance 1, the influence on the abnormality determination result is small. Therefore, the abnormality determination unit 170 selects a plurality of reference time series data x from the reference data Dref for the appliance operation period T3. (1) , x (2) , . . . , the accuracy of the abnormality determination can be improved while the calculation load on the abnormality determination unit 170 can be effectively reduced.

[0058] 12, each profile waveform PF is accompanied by a small sub-waveform SFsub. The sub-waveform SFsub has a smaller peak size than the main profile waveform PF, and therefore has less impact on abnormality determination. Therefore, by setting the equipment operation period T3 to include the profile waveform PF but not the sub-waveform SFsub, the calculation load on the abnormality determination unit 170 can be effectively reduced without reducing the accuracy of the abnormality determination.

[0059] Furthermore, in the abnormality determination device 100 according to another embodiment, when the device 1 has a plurality of operation modes and the characteristics of the state parameters change for each operation mode, the determination conditions serving as the basis for abnormality determination in the abnormality determination unit 170 may be switched based on the operation mode. For example, when the device 1 is a control valve used in a power plant facility as described above, and has a first operation mode M1 corresponding to a high load on the power plant facility and a second operation mode M2 ​​corresponding to a low load on the power plant facility, the determination conditions may be switched between the first operation mode M1 and the second operation mode M2.

[0060] Fig. 13 is a diagram showing changes over time in the state parameters in each operation mode, and Fig. 14 is a block diagram showing the functional configuration of an abnormality determination device 100 according to another embodiment. In Fig. 13, a first operation mode M1 corresponding to a high-load state is implemented up to time t0, and a second operation mode M2 ​​corresponding to a low-load state is implemented after time t0. In the second operation mode M2, the time width of the profile waveform increases compared to the first operation mode M1, indicating that the characteristics of the state parameter P of the device 1 change with respect to the load.

[0061] 14, the length ω of the sliding window may be used as a determination condition that can be switched between the first operation mode M1 and the second operation mode M2. For example, the length ω of the sliding window may be set as a determination condition in the first operation mode M1, in which the time width of the profile waveform PF is relatively small, while the length ω of the sliding window may be set as a determination condition in the second operation mode M2, in which the time width of the profile waveform PF is relatively large. As a result, even if the characteristics of the device 1 change when the operation mode is switched, an abnormality in the device can be suitably determined by changing the determination condition to correspond to the operation mode.

[0062] The present invention can be configured as an abnormality determination system 200 including a client terminal 190 capable of communicating with the abnormality determination device 100, as shown in FIG. The client terminal 190 is an information processing device having the hardware configuration shown in Fig. 3, and includes a request unit 200 as shown in Fig. 16. When an operation by an operator performed via the input unit 110 of the client terminal 190 is accepted, the request unit 200 requests the abnormality determination device 100 to execute the processing shown in the flowchart illustrating the abnormality determination method shown in Fig. 5, thereby enabling the abnormality determination device 100 to execute the processing. As described above, according to each of the above-described embodiments, time-series data of state parameters acquired from a device is compared with a plurality of reference time-series data extracted from reference data corresponding to the normal operating state of the device, thereby performing abnormality determination of the device. This allows appropriate abnormality determination to be performed based on a single time-series data.

[0063] In addition, within the scope of the present disclosure, the components in the above-described embodiments may be replaced with well-known components as appropriate, and the above-described embodiments may be combined as appropriate.

[0064] The contents described in each of the above embodiments can be understood, for example, as follows.

[0065] (1) An abnormality determination device according to one aspect includes: An abnormality determination device (100) for a device (1) in which a time change of a state parameter (P) during operation shows a specific profile waveform (PF), a time-series data acquisition unit (152) for acquiring time-series data (Dt) including a plurality of time-sequential data for the state parameter; a sliding window setting unit (180) that sets the length of a sliding window that indicates the number of data included in the time series data; From reference data (Dref) indicating the time change of the state parameter corresponding to the normal operating state of the device, a plurality of reference time series data (x (1) , x (2) an abnormality determination unit (170) for extracting the time series data having the length of the sliding window set by the sliding window setting unit and comparing the time series data having the length of the sliding window set by the sliding window setting unit with the plurality of reference time series data while sliding the sliding window, thereby determining whether or not an abnormality exists in the device; Equipped with.

[0066] According to the above aspect (1), time-series data of state parameters acquired from the device is compared with a plurality of reference time-series data extracted from reference data corresponding to the normal operating state of the device, thereby performing abnormality determination of the device, thereby making it possible to appropriately perform abnormality determination based on a single time-series data.

[0067] (2) In another embodiment, in the above embodiment (1), The abnormality determination unit identifies model data (Dm) from the plurality of reference time series data that includes at least a portion of the profile waveform, and determines whether or not there is an abnormality in the equipment by comparing the time series data having the length of the sliding window set by the sliding window setting unit with the model data while sliding the sliding window.

[0068] According to the above aspect (2), accurate abnormality determination can be performed by comparing the time series data acquired from the equipment with model data that includes at least a portion of a profile waveform that is likely to be affected when an abnormality occurs in the equipment.

[0069] (3) In another aspect, in the above aspect (1) or (2), The abnormality determination unit determines whether or not there is an abnormality in the equipment based on the degree of deviation (L) between the time series data and the plurality of reference time series data in a virtual space (V) having a dimension corresponding to the length of the sliding window set by the sliding window setting unit.

[0070] According to the above aspect (3), it is possible to suitably determine an abnormality in the device based on the degree of deviation between the time series data and a plurality of reference time series data in a virtual space of dimensions corresponding to the number of data included in the time series data.

[0071] (4) In another embodiment, in the above embodiment (3), The abnormality determination unit determines whether or not there is an abnormality in the device based on the degree of deviation between the time series data and the reference time series data that is closest to the kth time series data (k is any natural number) in the virtual space.

[0072] According to the above aspect (4), in the virtual space, it is possible to suitably determine an abnormality in the device based on the degree of deviation between the time series data and the k-th reference time series data among the plurality of reference time series data.

[0073] (5) In another embodiment, in any one of the above (1) to (4), The length (ω) of the sliding window is set so as to have a time width greater than at least one of the rising portion (PF1) and the falling portion (PF3) of the profile waveform.

[0074] According to the above aspect (5), in view of the fact that abnormalities occurring in equipment tend to be reflected in the rising and falling portions of the profile waveform, the sliding window length of the time series data is set to have a time width greater than at least one of the rising and falling portions of the profile waveform, thereby making it possible to suitably determine abnormalities in the equipment.

[0075] (6) In another embodiment, in the above embodiment (5), The length of the sliding window is set to have a time width greater than that of the profile waveform.

[0076] According to the above aspect (6), by setting the sliding window length of the time series data to have a time width larger than that of the profile waveform, it is possible to more appropriately determine an abnormality in the equipment.

[0077] (7) In another embodiment, in any one of the above (1) to (6), The time-series data acquisition unit acquires the time-series data during a data acquisition period (T1) that is defined to include an operation period of the device.

[0078] According to the above aspect (7), by acquiring time-series data including the operation period of the equipment, unnecessary data corresponding to the non-operation period of the equipment that makes little contribution to abnormality determination can be eliminated, thereby reducing the computational burden of abnormality determination and effectively improving the determination accuracy.

[0079] (8) In another embodiment, in the above embodiment (7), The abnormality determination unit extracts the plurality of reference time series data from the reference data during an equipment operation period (T3) including the profile waveform.

[0080] According to the above aspect (8), the plurality of reference time series data to be compared with the time series data acquired from the device are extracted from the device operation period including the profile waveform. This reduces the computational load for anomaly detection and effectively improves the detection accuracy by not extracting reference time series data from the reference data corresponding to the device non-operation period that contributes little to anomaly detection.

[0081] (9) In another embodiment, in any one of the above (1) to (8), The device has, as operation modes, a first operation mode (M1) corresponding to a high load and a second operation mode (M2) corresponding to a low load, The abnormality determination unit is capable of switching the abnormality determination conditions based on the operation mode.

[0082] According to the above aspect (9), the abnormality determination conditions are switched based on the operation mode. As a result, even if the characteristics of the device change due to the switching of the operation mode of the device, the abnormality of the device can be suitably determined by changing the determination conditions to correspond to the operation mode.

[0083] (10) In another embodiment, in the above embodiment (9), The determination conditions include the length of the sliding window set by the sliding window setting unit.

[0084] According to the above aspect (10), the number of data included in the time-series data is switched based on the operation mode.

[0085] (11) In another embodiment, in any one of the above (1) to (10), The device performs a cyclical operation.

[0086] According to the above aspect (11), a device that performs periodic operations is targeted for the determination. In such a device, a profile waveform appears periodically in the time variation of the state parameter used for the abnormality determination, so that the above configuration allows for appropriate abnormality determination.

[0087] (12) An abnormality determination method according to one aspect includes: A method for determining an abnormality in a device in which a time change of a state parameter (P) during operation shows a specific profile waveform (PF), comprising: an acquisition step of acquiring time series data (Dt) including a plurality of time-sequential data for the state parameter; a sliding window setting step of setting a length of a sliding window indicating the number of data included in the time series data; From reference data (Dref) indicating the time change of the state parameter corresponding to the normal operating state of the device, a plurality of reference time series data (x (1) , x (2) an abnormality determination step of extracting the time series data having the length of the sliding window set in the sliding window setting step and comparing the time series data having the length of the sliding window set in the sliding window setting step with the plurality of reference time series data while sliding the sliding window, thereby determining whether or not there is an abnormality in the device; Equipped with.

[0088] According to the above aspect (12), time-series data of state parameters acquired from the device is compared with a plurality of reference time-series data extracted from reference data corresponding to the normal operating state of the device, thereby performing abnormality determination of the device, thereby making it possible to appropriately perform abnormality determination based on a single time-series data.

[0089] (13) A program according to one aspect includes: Computer, an acquisition means for acquiring time-series data including a plurality of time-sequential data for a state parameter of a device whose time change in the state parameter exhibits a specific profile waveform during operation; a sliding window setting means for setting a length of a sliding window indicating the number of data included in the time series data; an abnormality determination means for extracting a plurality of reference time series data corresponding to the time series data from reference data indicating time changes of the state parameters corresponding to a normal operating state of the device, and for determining whether or not there is an abnormality in the device by comparing the time series data having the length of the sliding window set by the sliding window setting means with the plurality of reference time series data while sliding the sliding window; and make it work.

[0090] According to the above aspect (13), time-series data of state parameters acquired from the device is compared with a plurality of reference time-series data extracted from reference data corresponding to the normal operating state of the device, thereby performing abnormality determination of the device. This allows appropriate abnormality determination to be performed based on a single time-series data.

[0091] (14) An abnormality determination system according to one aspect includes: An abnormality determination system including an abnormality determination device for a device whose time-varying change in a state parameter during operation shows a specific profile waveform, and a client terminal, The abnormality determination device a time-series data acquisition unit for acquiring time-series data including a plurality of pieces of time-sequential data for the state parameter; a sliding window setting unit that sets a length of a sliding window that indicates the number of data included in the time series data; an abnormality determination unit that, in response to a request from the client terminal, extracts a plurality of reference time series data corresponding to the time series data from reference data that indicates time changes of the state parameters corresponding to a normal operating state of the device, and determines whether or not an abnormality exists in the device by comparing the time series data having the length of the sliding window set by the sliding window setting unit with the plurality of reference time series data while sliding the sliding window; Equipped with.

[0092] According to the above aspect (14), time-series data of state parameters acquired from the device is compared with a plurality of reference time-series data extracted from reference data corresponding to the normal operating state of the device in response to a request from the client terminal, thereby making it possible to appropriately perform abnormality determination based on a single time-series data.

[0093] (15) A control method for an abnormality determination system according to one aspect includes: A control method for an abnormality determination system including an abnormality determination device for a device whose time-varying state parameter during operation shows a specific profile waveform and a client terminal, comprising: The abnormality determination device a time-series data acquisition step for acquiring time-series data including a plurality of time-sequential data for the state parameter; a sliding window setting step of setting a length of a sliding window indicating the number of data included in the time series data; an abnormality determination step of extracting, in response to a request from the client terminal, a plurality of reference time series data corresponding to the time series data from reference data indicating time changes of the state parameters corresponding to a normal operating state of the device, and comparing the time series data having the length of the sliding window set in the sliding window setting step with the plurality of reference time series data while sliding the sliding window, thereby determining whether or not there is an abnormality in the device; Execute.

[0094] According to the above aspect (15), time-series data of state parameters acquired from the device is compared with a plurality of reference time-series data extracted from reference data corresponding to the normal operating state of the device in response to a request from the client terminal, thereby making it possible to appropriately perform abnormality determination based on a single time-series data. [Explanation of symbols]

[0095] 1 equipment 2 sensors 100 Abnormality determination device 110 Input section 120 Storage section 130 Arithmetic section 140 Output section 150 Data Monitoring Department 152 Time series data acquisition unit 160 Reference data storage unit 162 Reference time series data acquisition unit 170 Abnormality determination section 180 Sliding window setting section 190 client terminals 200 Request part Dref reference data Dt time series data L deviation degree M1 First operation mode M2 Second operating mode P state parameter PF profile waveform PF1 rising part PF2 peak PF3 Falling part SFsub Sub waveform T1 Data acquisition period T2 non-data acquisition period T3 Equipment operating period T4 Non-equipment operation period V Virtual Space

Claims

1. An abnormality determination device for a device in which a time change in a state parameter during operation shows a specific profile waveform, a time-series data acquisition unit for acquiring time-series data including a plurality of pieces of time-sequential data for the state parameter; a sliding window setting unit that sets a length of a sliding window that indicates the number of data included in the time series data; an abnormality determination unit for extracting a plurality of reference time series data corresponding to the time series data from reference data indicating time changes of the state parameter corresponding to a normal operating state of the device, and for determining whether or not there is an abnormality in the device by comparing the time series data having the length of the sliding window set by the sliding window setting unit with the plurality of reference time series data while sliding the sliding window; Equipped with The abnormality determination unit determines the presence or absence of an abnormality in the device based on the degree of deviation between a position corresponding to an ω-dimensional vector whose components are each consecutive data item included in the time series data and a position corresponding to an ω-dimensional vector whose components are each consecutive data item included in the plurality of reference time series data in a virtual space having an ω-dimensionality corresponding to the length of the sliding window set by the sliding window setting unit.

2. 2. The abnormality determination device according to claim 1, wherein the abnormality determination unit identifies model data including at least a portion of the profile waveform from among the plurality of reference time series data, and determines whether or not an abnormality exists in the device by comparing the time series data having the length of the sliding window set by the sliding window setting unit with the model data while sliding the sliding window.

3. 3. The abnormality determination device according to claim 1, wherein the abnormality determination unit determines whether or not an abnormality exists in the device based on a degree of deviation between the time series data and the reference time series data that is closest to the time series data by a kth order (k is an arbitrary natural number) in the virtual space.

4. The abnormality determination device according to claim 1 , wherein the length of the sliding window is set to have a time width greater than at least one of a rising edge and a falling edge of the profile waveform.

5. The abnormality determination device according to claim 4 , wherein the length of the sliding window is set to have a time width greater than that of the profile waveform.

6. The abnormality determination device according to claim 1 , wherein the time-series data acquisition unit acquires the time-series data during a data acquisition period that is defined to include an operation period of the device.

7. The abnormality determination device according to claim 6 , wherein the abnormality determination unit extracts the plurality of reference time-series data from the reference data during an equipment operation period including the profile waveform.

8. the device has, as operation modes, a first operation mode corresponding to a high load time and a second operation mode corresponding to a low load time; The abnormality determination device according to claim 1 , wherein the abnormality determination unit is capable of switching the abnormality determination conditions based on the operation mode.

9. The abnormality determination device according to claim 8 , wherein the determination condition includes a length of the sliding window set by the sliding window setting unit.

10. The abnormality determination device according to claim 1 , wherein the device performs a periodic operation.

11. A method for determining an abnormality in a device in which a time change in a state parameter during operation shows a specific profile waveform, comprising: an acquiring step of acquiring time-series data including a plurality of time-sequential data for the state parameter; a sliding window setting step of setting a length of a sliding window indicating the number of data included in the time series data; an abnormality determination step of extracting a plurality of reference time series data corresponding to the time series data from reference data indicating time changes of the state parameters corresponding to a normal operating state of the device, and determining whether or not there is an abnormality in the device by comparing the time series data having the length of the sliding window set in the sliding window setting step with the plurality of reference time series data while sliding the sliding window; Equipped with In the abnormality determination step, in a virtual space having an ω dimension corresponding to the length of the sliding window set in the sliding window setting step, the presence or absence of an abnormality in the device is determined based on the degree of deviation between a position corresponding to an ω-dimensional vector whose components are each consecutive data included in the time series data and a position corresponding to an ω-dimensional vector whose components are each consecutive data included in the plurality of reference time series data.

12. Computer, an acquisition means for acquiring time-series data including a plurality of time-sequential data for a state parameter of a device whose time change in the state parameter exhibits a specific profile waveform during operation; a sliding window setting means for setting a length of a sliding window indicating the number of data included in the time series data; an abnormality determination means for extracting a plurality of reference time series data corresponding to the time series data from reference data indicating time changes of the state parameters corresponding to a normal operating state of the device, and for determining whether or not there is an abnormality in the device by comparing the time series data having the length of the sliding window set by the sliding window setting means with the plurality of reference time series data while sliding the sliding window; and make it work, The abnormality determination means is a program for determining whether or not there is an abnormality in the equipment based on the degree of deviation between a position corresponding to an ω-dimensional vector whose components are each consecutive data included in the time series data and a position corresponding to an ω-dimensional vector whose components are each consecutive data included in the plurality of reference time series data, in a virtual space having an ω-dimension corresponding to the length of the sliding window set by the sliding window setting means.

13. An abnormality determination system including an abnormality determination device for a device whose time-varying change in a state parameter during operation shows a specific profile waveform, and a client terminal, The abnormality determination device a time-series data acquisition unit for acquiring time-series data including a plurality of pieces of time-sequential data for the state parameter; a sliding window setting unit that sets a length of a sliding window that indicates the number of data included in the time series data; an abnormality determination unit that, in response to a request from the client terminal, extracts a plurality of reference time series data corresponding to the time series data from reference data that indicates time changes of the state parameters corresponding to a normal operating state of the device, and determines whether or not an abnormality exists in the device by comparing the time series data having the length of the sliding window set by the sliding window setting unit with the plurality of reference time series data while sliding the sliding window; Equipped with the abnormality determination unit determines whether or not there is an abnormality in the device based on the degree of deviation between a position corresponding to an ω-dimensional vector whose components are each consecutive data item included in the time series data and a position corresponding to an ω-dimensional vector whose components are each consecutive data item included in the plurality of reference time series data in a virtual space having an ω-dimensionality corresponding to the length of the sliding window set by the sliding window setting unit.

14. A control method for an abnormality determination system including an abnormality determination device for a device whose time-varying state parameter during operation shows a specific profile waveform and a client terminal, comprising: The abnormality determination device a time-series data acquisition step for acquiring time-series data including a plurality of time-sequential data for the state parameter; a sliding window setting step of setting a length of a sliding window indicating the number of data included in the time series data; an abnormality determination step of extracting, in response to a request from the client terminal, a plurality of reference time series data corresponding to the time series data from reference data indicating time changes of the state parameters corresponding to a normal operating state of the device, and comparing the time series data having the length of the sliding window set in the sliding window setting step with the plurality of reference time series data while sliding the sliding window, thereby determining whether or not there is an abnormality in the device; Run a control method for an abnormality determination system, characterized in that in the abnormality determination step, in a virtual space having an ω dimension corresponding to the length of the sliding window set in the sliding window setting step, the presence or absence of an abnormality in the equipment is determined based on the degree of deviation between a position corresponding to an ω-dimensional vector whose components are each consecutive data included in the time series data and a position corresponding to an ω-dimensional vector whose components are each consecutive data included in the plurality of reference time series data.

15. An abnormality determination device for a device in which a time change in a state parameter during operation shows a specific profile waveform, a time-series data acquisition unit for acquiring time-series data including a plurality of pieces of time-sequential data for the state parameter; a sliding window setting unit that sets a length of a sliding window that indicates the number of data included in the time series data; an abnormality determination unit for extracting a plurality of reference time series data corresponding to the time series data from reference data indicating time changes of the state parameter corresponding to a normal operating state of the device, and for determining whether or not there is an abnormality in the device by comparing the time series data having the length of the sliding window set by the sliding window setting unit with the plurality of reference time series data while sliding the sliding window; Equipped with the device has, as operation modes, a first operation mode corresponding to a high load time and a second operation mode corresponding to a low load time; The abnormality determination unit is capable of switching the abnormality determination conditions based on the operation mode.

16. A method for determining an abnormality in a device in which a time change in a state parameter during operation shows a specific profile waveform, comprising: an acquiring step of acquiring time-series data including a plurality of time-sequential data for the state parameter; a sliding window setting step of setting a length of a sliding window indicating the number of data included in the time series data; an abnormality determination step of extracting a plurality of reference time series data corresponding to the time series data from reference data indicating time changes of the state parameters corresponding to a normal operating state of the device, and determining whether or not there is an abnormality in the device by comparing the time series data having the length of the sliding window set in the sliding window setting step with the plurality of reference time series data while sliding the sliding window; Equipped with the device has, as operation modes, a first operation mode corresponding to a high load time and a second operation mode corresponding to a low load time; In the abnormality determination step, the abnormality determination condition can be switched based on the operation mode.

17. Computer, an acquisition means for acquiring time-series data including a plurality of time-sequential data for a state parameter of a device whose time change in the state parameter exhibits a specific profile waveform during operation; a sliding window setting means for setting a length of a sliding window indicating the number of data included in the time series data; an abnormality determination means for extracting a plurality of reference time series data corresponding to the time series data from reference data indicating time changes of the state parameters corresponding to a normal operating state of the device, and for determining whether or not there is an abnormality in the device by comparing the time series data having the length of the sliding window set by the sliding window setting means with the plurality of reference time series data while sliding the sliding window; and make it work, the device has, as operation modes, a first operation mode corresponding to a high load time and a second operation mode corresponding to a low load time; The abnormality determination means is a program that can switch the abnormality determination conditions based on the operation mode.

Images