JP5485939B2 - Apparatus abnormality determination device and apparatus abnormality determination method - Google Patents

Description

  The present invention relates to a device abnormality determination device and a device abnormality determination method for determining whether or not a device constituting a facility is operating normally.

  An equipment abnormality generally refers to a state in which equipment constituting a facility (for example, a pump in a nuclear power plant) is not functioning normally. For example, there are a state where equipment parts are damaged and stopped, or a state where designed performance is not achieved. Possible causes of equipment abnormalities include disturbances, defects during manufacturing, construction and repair, deterioration over time due to wear and material deterioration, and overload during operation. A technique for obtaining operational data by attaching a sensor to a device and determining whether the device is operating normally from this operational data is referred to as device abnormality determination (or abnormality determination).

  The operation data of the device is sent to a server or the like from a plurality of sensors attached to the device and stored. For example, if the number of sensors is two, data of two sensor values at each measured time is sent to the server and stored in the server as device operation data.

  FIG. 2 shows an example of the data structure of the device operation data when the number of sensors is two. In the data structure shown in FIG. 2, the device operation data has two data of pressure and temperature as the sensor value 203 at each measured time 202.

  Data of sensor value 203 (pressure and temperature) at time 202 is sent from the sensor attached to the device to the server and stored in the server as device operation data. The sensor value 203 may not be acquired simultaneously for all sensors at the same time, but can be handled as a set of device operation data (data records) 201 within a certain time range (in FIG. 2 in units of one minute). .

  Abnormality determination is a technique for determining normal and abnormal binary values from a plurality of sensor values. As pre-processing for this binary determination, a value called the device abnormality level (a value obtained by quantifying the device abnormal state) is calculated from a plurality of sensor values, and whether the device abnormality level exceeds a threshold value is normal or abnormal. May be determined. In this case, the device abnormality degree is calculated for each set of device operation data acquired in time series.

FIG. 10A and FIG. 10B are examples of graphs showing the time change of the degree of device abnormality. As shown in FIGS. 10A and 10B, the time change of the device abnormality degree can be expressed by a graph in which the horizontal axis represents time and the vertical axis represents the device abnormality degree. A dotted line parallel to the horizontal axis indicates the threshold of the device abnormality degree. If the device abnormality level exceeds the threshold, it is determined that there is an abnormality. Thus, Figure 10A is abnormal at time _{t a,} FIG. 10B is determined to be normal at time _{t b.}

  However, even if the device abnormality level exceeds the threshold value, that is, even if it is determined to be abnormal, the device may not be stopped immediately. That is, even if the device abnormality level exceeds the threshold, there is a case that the device is not abnormal enough to stop the device. For example, this is the case when an abnormality is indicated by an external condition of the device.

  As an example, consider the case where the device is a pump. If the amount of water flowing into the pump is abnormally large, the sensor value may indicate an abnormal value, the device abnormality level may exceed a threshold value, and the abnormality determination may determine that there is an abnormality. However, in this case, it cannot be said that there is an abnormality in the pump itself. This device abnormality degree can be said to be a false abnormality degree caused by an abnormality outside the pump.

  In such a case, the abnormality determination is misreported. This is due to the fact that the state of the device is grasped only from the device operation data and that the amount of information is reduced because the device state is represented only by the value of the device abnormality level. More directly, it may be considered that the number of installed sensors and the installation location are not appropriate.

  Conversely, even if the device abnormality level is less than or equal to the threshold value, an early countermeasure may be necessary. For example, in the case of damage to a pump bearing, abnormalities may develop rapidly and damage the equipment unless they are dealt with early. In such a case, if the device abnormality is determined only by the threshold value, it is delayed to determine the abnormality. If an abnormality can be dealt with early, the damage to the parts will be limited and the maintenance cost will be kept low.

  As a case where determination cannot be made based on only the device abnormality level, a case where the device abnormality level vibrates in the vicinity of a threshold value can be considered. In response to this, for example, in Patent Document 1, after calculating the degree of abnormality, a contrivance is made such that a case where the degree of abnormality continues is regarded as abnormal. Moreover, in patent document 2, a failure possibility value is calculated by time-integrating the state quantity of an apparatus, and abnormality determination is performed.

JP 2008-37194 A JP 2010-25475 A

  However, in the techniques described in Patent Document 1 and Patent Document 2, if the state where the device abnormality degree exceeds the threshold value continues, it is determined as abnormal. For this reason, it may become a false report depending on the state of the device. Conversely, when the device abnormality level is equal to or less than the threshold value, it is not determined as an abnormality, and thus it is not possible to deal with a device abnormality even when an early countermeasure is required.

  In addition, with the techniques described in Patent Document 1 and Patent Document 2, it is impossible to give an appropriate instruction to maintenance personnel and operators in the actual site and to take actions to be taken in the event of an abnormality.

  The present invention has been made in view of the above points, and appropriately reflects the state of the equipment in the abnormality determination, reduces the false alarm of the abnormality determination, enables earlier abnormality determination, and is suitable for maintenance personnel and operators. It is an object of the present invention to provide a device abnormality determination device and a device abnormality determination method capable of taking corrective actions.

  The apparatus abnormality determination device according to the present invention is configured to receive a normal range database in which normal operation data representing an operation state at the normal time of a device is stored, and the measured operation state of the device is input as device operation data. An abnormality degree calculation unit that calculates an apparatus abnormality degree obtained by quantifying the degree of separation from the normal operation data, an abnormality degree database that stores the apparatus abnormality degree, and the apparatus abnormality degree at the time when the apparatus operation data is measured From each of the device abnormality degrees at a time that is a time interval that is set in advance of the measured time, a tendency calculation unit that obtains a time change in the device abnormality degree, and each of the fluctuation factors of the device abnormality degree, A fluctuation factor database storing a fluctuation range representing a magnitude of fluctuation of the device operation data from the normal operation data; and the device operation A variation contribution representing a time change of the data, a similarity is obtained for each of the variation factors from the variation contribution and the variation range, and the variation factor with respect to the maximum similarity is determined as the device abnormality degree. Corresponding behavior in which a variation factor estimation unit that estimates a variation factor, a result of abnormality determination and an action to be taken in the event of an abnormality are stored, which are determined by the device abnormality degree, the time variation of the device abnormality degree, and the variation factor A corresponding action determining unit that performs an abnormality determination by referring to the database and the corresponding action database and determining the degree of abnormality of the equipment and the time variation of the equipment abnormality degree and determining the action. .

  According to the present invention, it is possible to appropriately reflect the state of the device in the abnormality determination, to reduce the false report of the abnormality determination, and it is possible to determine the abnormality earlier than the conventional according to the variation factor of the device operation data. In addition, appropriate response actions can be presented to maintenance personnel and operators.

It is a block diagram which shows the structure of the apparatus abnormality determination apparatus by one example embodiment of this invention. It is an example of the data structure of apparatus operation data. It is a figure which shows an example of distribution of the apparatus operation data in a state space when the apparatus is operating normally. It is a figure which shows an example of the normal range in a state space. It is a figure which shows an example of the data structure of the normal operation data stored in normal range DB. It is a figure which shows an example of the data structure of variation factor DB. It is a processing flowchart which shows the calculation step which estimates the fluctuation | variation factor of an apparatus abnormality degree. It is a figure which shows an example of the data structure of corresponding | compatible action DB. It is a figure which shows an example of the data structure of abnormality degree DB. It is an example of a display of the result of abnormality determination and corresponding action. It is an example of a display of the result of abnormality determination and corresponding action. It is an example of a display of the result of abnormality determination and corresponding action. It is an example of the graph which shows the time change of an apparatus abnormality degree. It is an example of the graph which shows the time change of an apparatus abnormality degree. It is a figure which shows an example of the hardware constitutions of the apparatus abnormality determination apparatus by this invention.

Hereinafter, an embodiment of the present invention will be described in the following order with reference to the accompanying drawings.
1. 1. Configuration of device abnormality determination device Processing of Device Abnormality Determination Device Note that “device operation data” used in this specification is data including data (for example, temperature, pressure, etc.) indicating a state of operation of the device. Further, the “apparatus abnormality degree” is a value obtained by quantifying the abnormal state of the apparatus, and is a value representing a degree of separation indicating how far the input apparatus operation data is away from the normal range.

[1. Configuration of equipment abnormality determination device]
A configuration of a device abnormality determination device in an example of an embodiment of the present invention will be described with reference to FIG. FIG. 1 is a block diagram showing a configuration of a device abnormality determination device according to an embodiment of the present invention.

  The apparatus abnormality determination apparatus includes an input unit 101, an abnormality degree calculation unit 102, a tendency calculation unit 103, a fluctuation factor estimation unit 104, a corresponding action determination unit 105, and an output unit 106. As a database (DB), a normal range DB 110, An abnormality degree DB 113, a fluctuation factor DB 111, and a corresponding action DB 112 are provided.

  The input unit 101 is an interface for inputting device operation data. The device operation data is received from the outside of the apparatus and sent to the abnormality degree calculation unit 102. In addition to acquiring the device operation data, the device operation data may be processed and then sent to the abnormality degree calculation means 102 after acquisition. For example, when the device operation data is vibration data, since frequency analysis is effective, the frequency may be converted into intensity for each frequency by performing Fast Fourier Transform. If abnormal characteristics appear in a pulsed manner, it is effective to convert the number of pulses per unit time. For example, misfires such as engines and partial discharges such as motors and transformers correspond to this abnormality.

  The degree-of-abnormality calculation unit 102 calculates the degree of device abnormality using the acquired device operation data and the normal device operation data stored in the normal range DB 110. The calculation of the device abnormality level performed by the abnormality level calculation unit 102 will be described later.

  Hereinafter, the device operation data when the device is operating normally (when normal) is referred to as “normal operation data”.

  The normal range DB 110 is a database in which normal operation data is recorded, subjected to appropriate processing, and stored in advance. The data structure of the normal range DB 110 will be described later with reference to FIG.

  The abnormality degree DB 113 stores the calculated equipment abnormality degree together with corresponding equipment operation data (time and sensor value). The degree of abnormality DB 113 need not store all the calculated degree of equipment abnormality. It is sufficient to store only the device abnormality degree for the smallest time interval among the time intervals set in the trend calculation means 103 and the fluctuation factor estimation means 104. The data structure of the abnormality degree DB 113 will be described later with reference to FIG.

  The tendency calculation means 103 receives the calculated device abnormality degree, and obtains the tendency (time change) of the device abnormality degree from the inputted device abnormality degree and the past equipment abnormality degree extracted from the abnormality degree DB 113. Specifically, the device abnormality level at a time that is a predetermined time interval before the time corresponding to the input device abnormality level is extracted from the abnormality level DB 113, and the input device abnormality level and the abnormality level DB 113 are extracted. The difference between the extracted device abnormality degree is obtained and the inclination of the apparatus abnormality degree is calculated, and the apparatus abnormality degree is classified according to the inclination by attaching a trend label. The trend labels are, for example, “up”, “slow increase”, “small fluctuation”, and “down”. In the present embodiment, the trend labels have four levels, but the present invention is not limited to this. In addition, when the set time interval is constant, classification may be performed according to the difference, not the slope of the device abnormality degree. Further, instead of a simple slope, regression analysis may be performed to draw an approximate line of a high-order function such as a quadratic function or a function such as an exponential function, and the coefficient may be used.

  The variation factor estimation means 104 receives the calculated device abnormality degree, and estimates the variation factor of the device abnormality degree from the inputted device abnormality degree and the device operation data using the variation factor DB 111. Specifically, the device operation data at the time corresponding to the input device abnormality degree and the device operation data at the time preceding the time corresponding to the input device abnormality degree by a preset time interval. The difference is obtained, and the variation factor is estimated with reference to the variation factor DB 111. The time interval may be constant or may be changed according to environmental conditions such as time and temperature or the state of the device. For example, if the device abnormality level is relatively high, it is better to shorten the time interval.

  The variation factor DB 111 is a database that records what behavior the device operation data shows for each factor that causes the device abnormality degree to vary, and is created in advance. The variation factor DB 111 will be described later with reference to FIG.

  The corresponding action determining unit 105 refers to the corresponding action DB 112 based on the device abnormality degree, the tendency of the device abnormality degree (trend label), and the estimated fluctuation factor, and determines an appropriate corresponding action. The response action is an action that should be taken by a maintenance staff or an operator in the event of an abnormality, and is an action that deals with the abnormality.

  The corresponding action DB 112 is a database that is referred to when performing abnormality determination and determination of the corresponding action, and is created in advance. Corresponding behavior DB 112 stores device abnormality level, trend label, variation factor, abnormality determination result, and corresponding behavior, and corresponds to abnormality determination result corresponding to device abnormality level, trend label and variation factor. Action is defined. The response action DB 112 will be described later with reference to FIG.

  The output means 106 is means for arranging the obtained abnormality determination result and corresponding action on the screen and outputting them to the monitor. FIG. 9A, FIG. 9B, and FIG. 9C are output examples of abnormality determination results and corresponding actions. 9A to 9C will be described later.

  FIG. 11 is a diagram illustrating an example of a hardware configuration of the device abnormality determination device according to the present invention. The apparatus abnormality determination device includes a computer 1102 and a monitor 1103. The computer 1102 has a function of executing each unit described above and each DB described above. The monitor 1103 displays the result of abnormality determination and the corresponding action. The computer 1102 is connected to a device 1101 that is a target of abnormality determination through a signal line, and acquires a sensor value.

  Note that the device abnormality determination device may not include the monitor 1103. In this case, a separately prepared monitor (external monitor) is connected to the device abnormality determination device, and the result of abnormality determination and the corresponding action are displayed.

[2. Processing of equipment abnormality determination device]
Processing of the device abnormality determination device in an example of an embodiment of the present invention will be described. A pump is assumed as a device to be subjected to abnormality determination. Assume that a sensor for measuring pressure and temperature is attached to the pump, and device operation data including time, pressure and temperature as shown in FIG. 2 is obtained.

  First, the input unit 101 passes the device operation data to the abnormality degree calculation unit 102.

  Next, the abnormality degree calculation unit 102 calculates the device abnormality degree by comparing the normal operation data with the device operation data at the time of measurement (determination) received from the input unit 101. The calculated abnormality degree is stored in the abnormality degree DB 113.

  From here, a method for calculating the degree of device abnormality by the degree-of-abnormality calculation means 102 will be described.

  The pressure and temperature at the time t of the pump that is the target of the abnormality determination are considered to represent values representing the state of the pump at the time t. This value (pressure and temperature at time t) is called a data record at time t. For example, the data record 201 of the device operation data shown in FIG. 2 is considered to represent the state at the time 0:00 of the pump that is the target of abnormality determination. If this data record 201 is considered as a mathematical vector, the data record 201 is expressed as one point on a space having a dimension corresponding to the number of sensor values 203. This space is called a state space. In the case of the device operation data shown in FIG. 2, since the number of sensor values 203 included in the data record 201 is two, the state space is two-dimensional, that is, a plane.

  The degree of abnormality calculation means 102 calculates the degree of equipment abnormality using the state space. Hereinafter, a method for calculating the degree of device abnormality will be described with reference to FIGS. 3A, 3 </ b> B, and 4.

  FIG. 3A is a diagram illustrating an example of the distribution of device operation data in the state space when the device (pump) that is the target of abnormality determination is operating normally. The state space shown in FIG. 3A is a two-dimensional space of temperature and pressure. The device operation data at each time when the device is operating normally is usually gathered within a certain range when plotted in the state space. This range is called “normal range”. For example, a set of device operation data as shown in FIG. 3A is a normal range.

  However, if there is an abnormality such as a failure in the device, the device operation data deviates from the normal range and is plotted in the state space. Therefore, when the distribution range (normal range) of normal operation data is stored in advance, the degree of separation of how far the input device operation data (determination device operation data) is away from the normal range is calculated and obtained. The degree of separation can be regarded as the degree of equipment abnormality. In this case, since the scale differs for each sensor, normalization is required for each sensor. For example, it is conceivable to normalize the maximum amplitude to 1 around the average value.

  The normal range is a set of device operation data when the device is operating normally. The normal range is a range having a certain spread from each device operation data. This is because the device operation data does not represent all the possible states of the device, and therefore, the device state does not always correspond to any of the device operation data.

  FIG. 3B is a diagram illustrating an example of a normal range in the state space. The normal range shown in FIG. 3B is obtained by covering normal operation data with a plurality of circles (a supersphere in the state space if the state space is not two-dimensional). The centers and radii of the plurality of circles (hyperspheres) can be arbitrarily determined so as to include all normal operation data.

  In order to create a circle (hypersphere) for covering the device operation data in this way, for example, a technique called K-means clustering can be used. The following references introduce a method of dividing N data sets into K clusters by K-means clustering. A cluster is a data set in which the distance between points inside the cluster is smaller than the distance of data outside. To create a circle (hypersphere) as shown in FIG. 3B, the normal operation data is the data set in the reference, the center of the circle is the “prototype” of the cluster, and the radius of the circle is the farthest from the prototype The distance between the data and the prototype may be used.

  Reference: C.I. M.M. Bishop, Hiroshi Motoda, Takio Kurita, Tomoyuki Higuchi, Yuji Matsumoto, Noboru Murata, “Pattern Recognition and Machine Learning, Statistical Prediction Based on Bayesian Theory”, Springer Japan Co., Ltd., July 2008, p. 140-144.

  FIG. 4 is a diagram illustrating an example of the data structure of normal operation data stored in the normal range DB 110. The normal operation data is recorded in advance in the normal range DB 110.

  The normal operation data illustrated in FIG. 4 includes a plurality of records 401. One set of records 401 includes a circle name 402, a circle center (X, Y) 403, and a circle radius R 404. In FIG. 4, the circle name 402 is represented by Ck (k is 1 to n), the circle center 403 is represented by (Xk, Yk), and the circle radius 404 is represented by Rk. A distribution range (normal range) of normal operation data can be approximated by a set of circles represented by such records 401. The normal operation data only needs to have the circle center 403 and the circle radius 404, and the circle name 402 is not essential. In the example shown in FIG. 4, a circle name 402 is given for convenience of processing.

With reference to FIG. 3B, the degree-of-abnormality calculation unit 102 will be described. The degree-of-abnormality calculation unit 102 obtains a distance Dxk between the device operation data Px (Xa, Ya) obtained from the input unit 101 and the center (Xk, Yk) of each circle of normal operation data stored in the normal range DB 110. (In FIG. 3B, the case of k = 2 is shown). And according to Formula (1), among the things which remove | divided the distance Dxk by the radius Rk of each circle, the smallest thing is calculated as an apparatus abnormality degree.
Dxk = ((Xa−Xk) ² + (Ya−Yk) ² ) ^1/2
Device abnormality degree = min (Dxk / Rk | k = 1 to n) (1)
When the device abnormality degree is calculated in this way, when the threshold value is 1, and the device abnormality degree exceeds 1, it can be determined that the device is out of the normal operation data area (normal range).

  This is the end of the description of the device abnormality degree calculation method performed by the abnormality degree calculation unit 102.

  FIG. 8 is a diagram illustrating an example of the data structure of the abnormality degree DB 113. The calculated device abnormality degree is stored in the abnormality degree DB 113 together with the device operation data data record 201 shown in FIG. Therefore, the data record of the abnormality degree DB 113 includes time, sensor value (pressure and temperature), and equipment abnormality degree.

  Next, as described above, the trend calculation unit 103 compares the calculated device abnormality degree with the device abnormality degree at the previous time by a preset time interval, and calculates the inclination (time change) of the device abnormality degree. calculate. In the calculation, the data record of the abnormality degree DB 113 is used. For example, as shown in FIG. 8, when the time when the latest device abnormality degree is calculated is 0:10 and the set time interval is 10 minutes, the device abnormality degree at time 0:00 is determined from the abnormality degree DB 113. read out. Since the device abnormality level at time 0:10 is 1.2 and the device abnormality level at time 0:00 is 1.0, the slope of the device abnormality level is (1.2−1.0) / 10 minutes = 0.02 / min.

  Further, according to the determined inclination of the device abnormality degree, the four trend labels “up”, “slow increase”, “small fluctuation”, and “down” are classified. For example, when the slope is 0.07 or more, “rise”, when it is 0.025 or more and less than 0.07, “slowly rise”, when it is 0 or more and less than 0.025, “small fluctuation”, or less than 0 Is “down”.

  The fluctuation factor estimation means 104 obtains the contribution degree to the fluctuation of the equipment abnormality degree for each sensor value from the two data records of the abnormality degree DB 113 used by the tendency calculation means 103, and this contribution degree is determined as the data of the fluctuation factor DB 111. By comparing, the fluctuation factor of the device abnormality degree is estimated.

  FIG. 5 is a diagram illustrating an example of a data structure of the variation factor DB 111. The fluctuation factor DB 111 records what behavior the device operation data shows for each factor that changes the degree of device abnormality, and is created in advance. As shown in FIG. 5, the variation factor DB 111 includes a plurality of variation factor records 501, and the variation factor record 501 includes a variation factor name 502, a variation range 503 for each sensor value data constituting the device operation data, and the like. , And past cases 504.

  The variation factor name 502 represents a factor (variation factor) by which the degree of device abnormality varies.

  The fluctuation range 503 represents the magnitude of the fluctuation of how much the sensor value fluctuates from the normal range with respect to each fluctuation factor, and the displacement (dX from the normal range in the state space shown in FIG. 3B. , DY). The fluctuation range 503 is normalized so that the magnitude becomes 1 when this is regarded as a vector.

  The variation factor DB 111 can be created by analyzing data when a device failure actually occurs or by performing a physical simulation.

  In the example shown in FIG. 5, the past case 504 is included in the items of the variation factor record 501. The past case 504 is information on the state and state of the device when the variation described in the variation factor name 502 has occurred in the past, and abnormally such as how the device became after the variation occurred. Information useful to deal with can be included. In FIG. 5, for example, for input fluctuations, it is described that the degree of abnormality has decreased after several days. The past case 504 may be described for all the variable factor names 502 or only for a specific variable factor name 502.

  In the past case 504, information on the device itself that is the target of the abnormality determination is desirable, but devices of the same model or devices having the same operation principle may be referred to. In addition, it is more useful to include not only one piece of information but also a plurality of pieces of information in the past case 504. In this case, a past case database in which a plurality of past cases are recorded may be mounted and associated with these past cases.

  FIG. 6 is a process flow diagram showing calculation steps for estimating the variation factor of the device abnormality degree, which is performed by the variation factor estimation means 104.

In step 601, the magnitude of variation between data records is calculated. For example, in the example of FIG. 8 described above, sensor values (pressure and temperature) at time 0:00 and time 0:10 are read from the abnormality degree DB 113. The sensor value at time 0:00 is (Xt1, Yt1), and the sensor value at time 0:10 is (Xt2, Yt2). And the difference of these sensor values is calculated by Formula (2).
Sensor value difference = ((Xt2-Xt1), (Yt2-Yt1)) (2)
In equation (2), the sensor value is regarded as a vector.

In step 602, the difference value Dt2 of the sensor values is calculated by the equation (3), and the contribution of each sensor value to the variation factor, that is, the variation contribution (dXt2, dYt2) is calculated by the equation (4).
Dt2 = ((Xt2-Xt1) ² + (Yt2-Yt1) ² ) ^1/2 (3)
Fluctuation contribution (dXt2, dYt2) = ((Xt2-Xt1) / Dt2, (Yt2-Yt1) / Dt2) (4)
The variation contribution degree (dXt2, dYt2) is a value obtained by dividing the difference between the sensor values by the magnitude Dt2 of the difference between the sensor values, and represents a time change of each sensor value.

In step 603, the degree of similarity between the variation contribution and each variation factor record 501 in the variation factor DB 111 is calculated. The similarity St2 is calculated by the equation (5). The similarity St2 corresponds to an inner product when the variation contribution (dXt2, dYt2) and each variation width (dX, dY) 503 of the variation factor record 501 are regarded as vectors.
Similarity St2 = (dXt2 × dX + dYt2 × dY) (5)
The similarity St2 is calculated for all the variable factor records 501. That is, in equation (5), (dX, dY) is replaced with all of (dXa, dYa), (dXb, dYb), (dXc, dYc)... Shown in FIG. St2 is calculated.

  In step 604, the largest similarity among the similarities calculated in step 603 is obtained, and a variation factor for the maximum similarity is set as a variation factor of the device abnormality degree.

  After estimating the variation factor of the degree of device abnormality, the variation factor estimation unit 104 reads the past case 504 corresponding to this variation factor from the variation factor DB 111.

  Next, the corresponding action determining unit 105 determines abnormality using the device abnormality degree calculated by the abnormality degree calculating unit 102, the tendency label determined by the tendency calculating unit 103, and the variation factor estimated by the variation factor estimating unit 104. To determine the response action. The abnormality determination and the response action determination are performed with reference to the response action DB 112.

  FIG. 7 is a diagram illustrating an example of the data structure of the corresponding action DB 112. The correspondence action DB 112 shown in FIG. 7 has three types of items, that is, a device abnormality degree 701, a tendency label 702, and a fluctuation factor 703, and the abnormality determination result and the corresponding action determined according to these items. Stored. In FIG. 7, the abnormality determination result and the corresponding action are described in cells (cells in each column of the variation factor 703) classified by the device abnormality degree 701, the trend label 702, and the variation factor 703. In the cell, the result of abnormality determination is described in the upper part, and the corresponding action is described in the lower part. The corresponding action is described as “(no display)” when it is not necessary to display it.

  The result of the abnormality determination and the corresponding action can be determined depending on whether the device abnormality degree 701 is equal to or greater than the threshold value or less than the threshold value. In the present embodiment, the device abnormality degree 701 is calculated using the equation (1) as described above, and thus the threshold value can be set to 1. Therefore, whether or not the device abnormality level 701 is 1 or more is one of the criteria for determining the result of abnormality determination and the corresponding action.

  The tendency label 702 represents the tendency of the equipment abnormality level, that is, the progress degree of the equipment abnormality degree. The value of the tendency label 702 is one of the criteria for determining the abnormality determination result and the corresponding action.

  The variation factor 703 is a variation factor estimated by the variation factor estimation unit 104, and this is also one of the criteria for determining the result of the abnormality determination and the corresponding action.

  The corresponding action determining unit 105 refers to the corresponding action DB 112 and reads out the abnormality determination result and the corresponding action described in the corresponding cell from the device abnormality degree 701, the tendency label 702, and the variation factor 703. For example, when the device abnormality degree 701 is less than the threshold, the trend label 702 is “slowly rising”, and the variation factor 703 is “bearing damage”, the abnormality determination result is “abnormal”, and the corresponding action is “necessary”. Attention monitoring ”.

  Next, the output unit 106 displays the result of the abnormality determination and the corresponding action on the monitor screen.

  FIG. 9A, FIG. 9B, and FIG. 9C are display examples of abnormality determination results and corresponding actions. On the screen, the name of the device that is the target of the abnormality determination is displayed as the target 801, and a graph representing the time change of the device abnormality level obtained from the abnormality level DB 113 is displayed below the name of the device (target 801). 809 is displayed. In the graph 809, the horizontal axis represents time, the vertical axis represents the device abnormality level, the device abnormality degree threshold value 802 is indicated by a dotted line, and a point subjected to abnormality determination is indicated by a triangle mark 803.

  On the right side of the graph 809, the abnormality determination result acquired by the corresponding action determining unit 105 is displayed as a state 804. Below the state 804, a variation factor 805 of the degree of equipment abnormality estimated by the variation factor estimation means 104 is displayed. Below the fluctuation factor 805, the trend label obtained by the trend calculation unit 103 is displayed as an abnormal progress trend 806. Under the progress tendency 806, the corresponding action 807 acquired by the corresponding action determining unit 105 is displayed. A past case 808 obtained by the variation factor estimating means 104 is displayed below the corresponding action 807.

  Note that the output means 106 includes a graph 809 representing a time change in the degree of device abnormality, a state 804 as a result of abnormality determination, a fluctuation factor 805, an abnormality progress tendency 806, a corresponding action 807, and past cases 808. Of these, at least one of them may be output.

  In FIG. 9A, the degree of device abnormality is equal to or greater than the threshold value, but the fluctuation factor 805 is “input fluctuation” and the progress tendency 806 (trend label) is “rising”, and therefore, based on the corresponding action DB 112 shown in FIG. The state 804 that is the result of the abnormality determination is “normal”, and the corresponding action 807 is “caution”. Thus, even if the device abnormality level is equal to or greater than the threshold value, the abnormality determination result may be “normal”. The past case 808 is “decreasing degree of abnormality after several days” from the variation factor DB 111 shown in FIG.

  In FIG. 9B, the degree of device abnormality is less than the threshold, but the fluctuation factor 805 is “bearing damage” and the progress tendency 806 (trend label) is “slowly rising”, so the corresponding action DB 112 shown in FIG. Based on the above, the state 804 as the result of the abnormality determination is “abnormal”, and the corresponding action 807 is “monitoring that requires attention”. Thus, even if the device abnormality level is less than the threshold value, the result of the abnormality determination may be “abnormal”. The past case 808 becomes “abrupt failure” from the variation factor DB 111 shown in FIG.

  In FIG. 9C, the degree of device abnormality is equal to or greater than the threshold, but the fluctuation factor 805 is “seal wear” and the progress tendency 806 (trend label) is “slowly rising”, so the corresponding action DB 112 shown in FIG. Based on the above, the state 804 that is the result of the abnormality determination is “normal”. The corresponding action 807 is not displayed because “(not displayed)” is described in the corresponding action DB 112. The past case 808 is not displayed because it is not described in the variation factor DB 111 shown in FIG. As described above, even if the device abnormality level is equal to or higher than the threshold value, the corresponding action 807 and the past case 808 may not be displayed.

  As described above, the device abnormality determination apparatus according to the present invention performs abnormality determination based on the device abnormality degree, the tendency (time change) of the device abnormality degree, and the fluctuation factor of the device abnormality degree. There are fewer false alarms than the judgment device. Even if the device abnormality level is less than the threshold value, it is determined that there is an abnormality according to the tendency and variation factors of the device abnormality level, so that it is possible to cope with the abnormality early. Since the apparatus abnormality determination device according to the present invention displays the corresponding action based on the abnormality determination as shown in FIGS. 9A and 9B, the maintenance staff and the operator can take an appropriate corresponding action.

  DESCRIPTION OF SYMBOLS 101 ... Input means, 102 ... Abnormality calculation means, 103 ... Trend calculation means, 104 ... Fluctuation factor estimation means, 105 ... Corresponding action determination means, 106 ... Output means, 110 ... Normal range DB, 111 ... Fluctuation factor DB, 112 ... corresponding action DB, 113 ... abnormality degree DB, 201 ... data record, 202 ... time, 203 ... sensor value, 401 ... record, 402 ... name of circle, 403 ... center of circle, 404 ... radius of circle, 501 ... fluctuation Factor record, 502 ... variation factor name, 503 ... variation range, 504 ... past case, 701 ... device abnormality level, 702 ... trend label, 703 ... variation factor, 801 ... target (name of device), 802 ... device abnormality degree Threshold value 803 ... Triangular mark indicating a point to be subjected to abnormality determination, 804 ... State, 805 ... Variation factor, 806 ... Progression tendency, 807 ... Corresponding action, 808 ... Past events , 809 ... graph, 1101 ... abnormality determination subject to equipment, 1102 ... computer, 1103 ... monitor.

Claims (12)

A normal range database in which normal operation data representing the normal operation status of the device is stored;
The measured operation state of the device is input as device operation data, and an abnormality degree calculation unit that calculates a device abnormality degree that quantifies the degree of separation of the device operation data from the normal operation data,
An abnormality degree database for storing the device abnormality degree;
Trend calculation for determining a time change of the device abnormality degree from the device abnormality degree at the time when the device operation data is measured and the device abnormality degree at a time before the measured time by a preset time interval. And
For each of the fluctuation factors of the equipment abnormality degree, a fluctuation factor database in which a fluctuation range representing a magnitude of fluctuation of the equipment operation data from the normal operation data is stored;
A variation contribution representing a time change of the device operation data is obtained, a similarity is obtained for each of the variation factors from the variation contribution and the variation range, and the variation factor with respect to the maximum similarity is determined as the device abnormality. A fluctuation factor estimator that estimates the fluctuation factor of the degree,
A corresponding behavior database that stores the result of abnormality determination and the action to be taken in the event of an abnormality, defined by the device abnormality degree, the time variation of the equipment abnormality degree, and the variation factor;
With reference to the corresponding behavior database, from the variation factor estimated as the device abnormality degree and the time variation of the device abnormality degree, a corresponding behavior determination unit that performs abnormality determination and determines the behavior;
An apparatus abnormality determination device comprising:   The apparatus abnormality determination device according to claim 1, wherein the trend calculation unit classifies the apparatus abnormality degree according to a magnitude of a time change.   The past example that is information about the device when a variation has occurred in the past due to the variation factor is stored in the variation factor database for at least one of the variation factors of the device abnormality degree. Equipment abnormality judgment device.   The apparatus abnormality determination apparatus according to claim 1, further comprising: an output unit that outputs at least one of the time variation of the apparatus abnormality degree, the estimated variation factor, the result of the abnormality determination, and the behavior.   The output part which outputs at least 1 among the time change of the said apparatus abnormality degree, the estimated said fluctuation factor, the result of the said abnormality determination, the said action, and the classification | category by the said classification is provided. Equipment abnormality judgment device.   The apparatus abnormality of Claim 3 provided with the output part which outputs at least 1 among the time change of the said apparatus abnormality degree, the estimated said fluctuation factor, the result of the said abnormality determination, the said action, and the said past example. Judgment device. In the device abnormality determination method for determining abnormality of the device based on the measured operating state of the device using a computer,
The abnormality degree calculation unit of the computer inputs the operation state of the device as device operation data, refers to a normal range database in which normal operation data representing the operation state of the device at normal time is stored, and the normal operation An abnormality degree calculating step for calculating an apparatus abnormality degree obtained by quantifying the degree of separation of the apparatus operation data from data,
The tendency calculation unit of the computer refers to the abnormality degree database in which the degree of abnormality of the device is stored, and the degree of abnormality of the device at the time when the device operation data is measured, and a time interval set in advance from the time of measurement. From the device abnormality degree at a time just before, a trend calculation step for obtaining a time change of the device abnormality degree;
The fluctuation factor estimation unit of the computer refers to a fluctuation factor database in which a fluctuation range representing a magnitude of fluctuation of the device operation data from the normal operation data is stored for each of the fluctuation factors of the device abnormality degree, A variation contribution representing a time change of the device operation data is obtained, a similarity is obtained for each of the variation factors from the variation contribution and the variation range, and the variation factor with respect to the maximum similarity is determined as the device abnormality. A fluctuation factor estimation step for estimating a fluctuation factor of the degree,
A correspondence behavior database in which the correspondence behavior determination unit of the computer stores a result of abnormality determination and an action to be taken when an abnormality occurs, which is determined by the degree of abnormality of the equipment, a time change of the degree of equipment abnormality, and the variation factor; Referring to the degree of equipment abnormality and the time variation of the degree of equipment abnormality and the estimated variation factor, making a judgment of abnormality and determining the corresponding action,
A device abnormality determination method comprising:   The apparatus abnormality determination method according to claim 7, wherein in the trend calculation step, the trend calculation unit classifies the degree of apparatus abnormality according to a magnitude of a time change.   In the variation factor estimation step, the variation factor estimation unit stores a past case, which is information about the device when a variation has occurred in the past due to the variation factor, for at least one of the variation factors of the device abnormality degree. The apparatus abnormality determination method according to claim 7, wherein the fluctuation factor database is referred to.   The output part of the said computer is provided with the output step which outputs at least 1 among the time change of the said apparatus abnormality degree, the estimated said fluctuation factor, the result of the said abnormality determination, and the said action. Device abnormality judgment method.   An output step in which the output unit of the computer outputs at least one of the time change of the device abnormality degree, the estimated variation factor, the result of the abnormality determination, the behavior, and the classification by the classification. The apparatus abnormality determination method of Claim 8 provided with.   The output unit of the computer includes an output step of outputting at least one of the time change of the device abnormality degree, the estimated variation factor, the result of the abnormality determination, the behavior, and the past case. The apparatus abnormality determination method according to claim 9.

Images