Device status monitoring apparatus and device status monitoring method

Abstract

The device state monitoring device (1) projects a plurality of measured data representing the state value of the device into a dimensionless space, and estimates the distribution of the state of the device based on the plurality of measured data projected into the dimensionless space, wherein the dimensionless space represents a plurality of display shapes with a common shape, and the plurality of display shapes respectively represent a plurality of normal information representing the normal state of the device. Furthermore, based on the relationship between the measured data representing the normal state of the device in the dimensionless space and the display shapes representing the normal information, both or either of the normal information and the measured data in the dimensionless space are corrected.

Description

Technical Field

[0001] The present invention relates to a device state monitoring apparatus and a device state monitoring method for monitoring the state of a device. Background Art

[0002] As a prior art for monitoring the state of equipment, there is a maintenance period determination method for a vehicle air conditioner described in Patent Document 1. In this method, the difference between the specific enthalpy at the start of compression and the specific enthalpy at the end of compression of the outdoor heat exchanger is compared with a first set value and a second set value greater than the first set value, and when the difference in specific enthalpy is greater than the second set value, the air conditioner is determined to require maintenance.

[0003] On the other hand, when the difference in specific enthalpy is greater than the first set value and less than the second set value, a refrigeration cycle diagram of a Mollier diagram (hereinafter referred to as a ph (pressure-specific enthalpy) diagram) is generated based on sensor data detected by various sensors installed in the outdoor heat exchanger. This refrigeration cycle diagram is displayed together with a refrigeration cycle diagram when the air conditioner is in a normal state.

[0004] In the ph line diagram, when the shape of the refrigeration cycle graph changes from the shape corresponding to the normal state, it can be determined that the performance of the air conditioner has deteriorated. When the difference in the shape exceeds a set value, the air conditioner is determined to require maintenance. In addition, the refrigeration cycle graph in the normal state of the air conditioner is a display shape of normal information indicating the normal state of the equipment, and the sensor data used to generate the refrigeration cycle graph is the actual measured data of the state value of the equipment.

[0005] Prior art literature

[0006] Patent Literature

[0007] Patent Document 1: Japanese Patent Application Publication No. 2015-92121 Summary of the invention

[0008] Problems to be solved by the invention

[0009] The shape of the refrigeration cycle graph of the pH diagram of the air conditioner is offset according to the outside air temperature even if the set temperature of the air conditioner is the same. That is, the refrigeration cycle graph in the normal state of the air conditioner has a different shape according to the outside air temperature. In addition, the refrigeration cycle graph compared with the refrigeration cycle graph in the normal state of the air conditioner is the so-called instantaneous data generated based on the sensor data detected at the start time point of the maintenance period determination.

[0010] In the method described in Patent Document 1, it is determined whether the equipment represented by the instantaneous data is in a normal state by monitoring whether the equipment needs maintenance. Therefore, when the instantaneous data is an outlier generated by chance, the state of the equipment may be erroneously identified. The influence of the outlier generated by chance in the sensor data can be reduced by the change tendency of the state of the equipment determined by using a plurality of sensor data detected sequentially within a certain period (e.g., within 1 month).

[0011] However, within a certain period of time, when the external environment of the device changes or the method of using the device is changed, the normal state of the device deviates according to these changes. Therefore, the normal information of the device is represented by a plurality of display shapes that are different according to the deviation of the normal state of the device. Therefore, it is not easy to know the correspondence between the plurality of sensor data and the normal information represented by the plurality of display shapes, and there is a problem that the state of the device cannot be accurately identified.

[0012] Furthermore, in the method described in Patent Document 1, a physical model simulating the physical state of the air conditioner is used to set normal information for each operating state of the air conditioner (for example, a refrigeration cycle diagram of the normal operating state of the air conditioner). The physical model simulates the physical state of the device operating under a preset ideal operating environment, and therefore, by using the physical model, the normal information of the device operating under the ideal operating environment is calculated.

[0013] Generally speaking, the ideal operating environment of a device is different from the actual operating environment of the device, so an error occurs in the normal information of the device in the actual operating environment and the normal information assuming the ideal operating environment of the device. Therefore, there is an offset corresponding to the error between the sensor data detected from the device in the actual operating environment and the normal information assuming the ideal operating environment of the device, so there is a problem that the state of the device cannot be accurately identified even if they are compared.

[0014] The present invention is to solve the above-mentioned problems, and an object of the present invention is to obtain a device state monitoring apparatus and a device state monitoring method capable of accurately identifying the state of a device.

[0015] Means for solving problems

[0016] The device status monitoring device of the present invention comprises: a data acquisition unit, which acquires measured data representing the status value of the device; a normal information acquisition unit, which acquires normal information representing the normal status of the device; a position relationship acquisition unit, which acquires the positional relationship of the measured data in the display shape representing the normal information of the device; a projection unit, which projects the multiple measured data into a dimensionless space based on the positional relationship between the multiple display shapes representing the multiple normal information of the device and the multiple measured data, and the dimensionless space uses a common shape to represent the multiple display shapes representing the multiple normal information; a correction unit, which corrects both or either one of the normal information and the measured data in the dimensionless space based on the relationship between the measured data representing the normal status of the device and the display shape representing the normal information in the dimensionless space; and a distribution estimation unit, which estimates the distribution of the status of the device based on the multiple measured data projected into the dimensionless space.

[0017] Effects of the Invention

[0018] According to the present invention, a plurality of measured data representing the state value of the device are projected into a dimensionless space, and the distribution of the state of the device is estimated based on the plurality of measured data projected into the dimensionless space. The dimensionless space represents a plurality of display shapes with a common shape, and the plurality of display shapes respectively show a plurality of normal information representing the normal state of the device. Furthermore, based on the relationship between the measured data representing the normal state of the device in the dimensionless space and the display shape representing the normal information, both or either of the normal information and the measured data in the dimensionless space are corrected. Thus, the device state monitoring device of the present invention can accurately identify the state of the device. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] [ Figure 1 ] is a block diagram showing the structure of the device status monitoring device according to embodiment 1.

[0020] [ Figure 2 ] is a flowchart showing the device status monitoring method of implementation mode 1.

[0021] [ Figure 3 ] Figure 3 A is a graph showing the relationship between a plurality of display shapes respectively representing a plurality of normal information of a device and a plurality of sensor data of the device, Figure 3 B is a graph showing a dimensionless space into which multiple sensor data of a device are projected.

[0022] [ Figure 4 ] is an explanatory diagram showing an overview of the correction processing of the normal information of the device.

[0023] [ Figure 5 ] is a flowchart showing the positional relationship calculation process.

[0024] [ Figure 6 ] is a graph showing the relationship between a display shape representing normal information of a device and sensor data detected from the device.

[0025] [ Figure 7 ] is a graph showing a display shape representing normal information of a device, sensor data detected from the device, and a center point of the display shape representing normal information of the device.

[0026] [ Figure 8 ] is a graph showing the positional relationship between the display shape of normal information representing a device and the sensor data detected from the device.

[0027] [ Fig. 9 ] is a flow chart showing the projection process of sensor data onto a dimensionless space.

[0028] [ Fig.10 ] is a graph showing an example of a dimensionless space.

[0029] [ Fig.11 ] is shown projected onto Fig.10 A plot of sensor data in dimensionless space.

[0030] [ Fig.12 ] is a graph showing an overview of the correction processing of sensor data of the device.

[0031] [ Fig.13 ] is a flowchart showing the projection processing and correction processing of sensor data of the device.

[0032] [ Fig.14 ] is a graph showing the positional relationship of sensor data of devices in a dimensionless space.

[0033] [ Fig.15 ] Fig.15 A is a graph showing the distribution of sensor data projected into a dimensionless space, Fig.15 B is an explanatory diagram showing an overview of a process of estimating the state of a device based on the distribution of sensor data projected into a dimensionless space.

[0034] [ Fig.16 ] Fig.16 A is a block diagram showing a hardware configuration that realizes the functions of the device state monitoring device of Embodiment 1. Fig.16 B is a block diagram showing a hardware configuration for executing software that realizes the functions of the device state monitoring device according to the first embodiment. DETAILED DESCRIPTION

[0035] Implementation Method 1

[0036] Figure 1 1 is a block diagram showing the structure of the device state monitoring device 1 according to the first embodiment. Figure 1 In the present invention, the device state monitoring device 1 is a device that monitors the state of the device based on the result obtained by comparing the measured data representing the state value of the monitored device with the display shape of the normal information representing the device in the dimensionless space. The normal information representing the normal state of the device is, for example, the range of the state value representing the normal operation state of the device, which is obtained according to each operation state of the device.

[0037] like Figure 1 As shown, the device state monitoring device 1 includes a data acquisition unit 11, a normal information acquisition unit 12, a positional relationship acquisition unit 13, a projection unit 14, a correction unit 15, a distribution estimation unit 16, and an output unit 17. A sensor for detecting various operating state values ​​of the device is provided in the device whose operating state is monitored by the device state monitoring device 1. The monitored device is, for example, an air conditioner.

[0038] The data acquisition unit 11 acquires measured data representing the status value of the monitored device. For example, the data acquisition unit 11 acquires measured data for a certain period of time during which the status of the device is monitored. The certain period of time during which the status of the device is monitored is, for example, about one month. The measured data is, for example, sensor data detected by sensors installed in the device. In the case where the device is an air conditioner, the sensors are, for example, pressure sensors and temperature sensors installed in various parts of the air conditioner. Based on the sensor data detected by these sensors, normal information of the air conditioner is determined according to each operating state of the air conditioner.

[0039] The normal information acquisition unit 12 acquires normal information of the monitored device. When the monitored device is an air conditioner, the normal information is, for example, a refrigeration cycle of a ph line diagram of the air conditioner in a normal state. For example, the normal information acquisition unit 12 calculates the refrigeration cycle of the air conditioner in a normal operation state using sensor data in a normal operation state detected by a pressure sensor and a temperature sensor provided in the air conditioner. That is, in the "acquisition" of normal information by the normal information acquisition unit 12, in addition to reading and acquiring the normal information stored in the storage device, it also includes the case where the normal information is calculated and acquired using the measured data representing the state value of the device.

[0040] The display shape representing the normal information of the air conditioner is, for example, a refrigeration cycle diagram in the normal operating state of the air conditioner. In addition, the refrigeration cycle diagram of the air conditioner changes according to the external air temperature, and there are multiple shapes corresponding to various external air temperatures in the refrigeration cycle diagram representing the normal information of the air conditioner. In addition, the sensor provided in the air conditioner detects multiple sensor data for each operating state of the air conditioner. Therefore, when simply comparing multiple sensor data and normal information represented by multiple display shapes, it is not easy to know the corresponding relationship between the sensor data distribution and the normal information, and it is impossible to accurately identify the state of the device.

[0041] The positional relationship acquisition unit 13 acquires the positional relationship of the measured data in the display shape of the normal information of the device. For example, when the display shape representing the normal information of the device is a polygon, the positional relationship acquisition unit 13 calculates the distance between the center point of the polygon and the measured data as the positional relationship of the measured data in the polygon, and calculates the angle between the straight line passing through the center point of the polygon and the measured data and the side of the polygon intersecting the straight line. The center point of the polygon is, for example, the centroid.

[0042] Furthermore, “acquisition” of the positional relationship by the positional relationship acquisition unit 13 includes not only calculation of the positional relationship using the actual measurement data and the normal information, but also reading and acquiring the positional relationship data stored in the storage device.

[0043] The projection unit 14 projects the plurality of measured data onto a dimensionless space based on the positional relationship between the plurality of display shapes representing the plurality of normal information of the device and the plurality of measured data. The dimensionless space is a space where the plurality of display shapes representing the plurality of normal information are represented by a common shape, and plot points (dimensionless) representing the positions of the measured data in the display shapes representing the normal information are arranged (projected).

[0044] The common shape is, for example, a circle with a radius of 1. The projection unit 14 applies the center point of the display shape of multiple normal information corresponding to the multiple measured data to the center point of the circle, and converts the positional relationship between the center point of the display shape of the normal information and the measured data into a positional relationship with the center point of the circle. The positional relationship is, for example, distance information and angle information. The projection unit 14 configures multiple plotted points of measured data in a dimensionless space including a circle. The plotted points of the measured data configured in the dimensionless space are points of dimensionless values ​​representing the positional relationship of the measured data relative to the display shape of the normal information representing the device. In addition, the distribution of the plotted points in the dimensionless space is equivalent to the distribution of the plotted points corresponding to the state values ​​of the device.

[0045] The correction unit 15 corrects both or either one of the normal information and the measured data in the dimensionless space according to the relationship between the measured data representing the normal state of the device in the dimensionless space and the display shape representing the normal information. For example, when the display shape representing the normal information of the device is a circle with a radius of 1 in the dimensionless space, the circle serves as a threshold for determining whether the operating state of the device is a normal state or an abnormal state. In addition, the measured data detected from the device in the normal operating state is mostly distributed at the center of the circle with a radius of 1 representing the range of normal state values ​​of the device. The center of the circle is equivalent to the origin of the coordinate system set in the dimensionless space (hereinafter referred to as the origin of the dimensionless space). In other words, the offset between the center position of the distribution of the measured data projected into the dimensionless space and the center position of the circle with a radius of 1 is equivalent to the error between the normal information in the actual operating environment and the normal information assuming an ideal operating environment.

[0046] Therefore, the correction unit 15 calculates the shortest distance between the center position of the distribution of the measured data projected into the dimensionless space and a circle with a radius of 1, and corrects the normal information and the measured data so that the display shape representing the normal information becomes a circle with the calculated shortest distance as the radius. Thus, the relationship between the normal information and the measured data of the device is corrected, and by comparing them, the state of the device can be accurately identified.

[0047] The distribution estimation unit 16 estimates the distribution of the state of the device based on the plurality of measured data projected into the dimensionless space. For example, the distribution estimation unit 16 performs a mixed Gaussian model estimation process on the plotted points of the plurality of measured data arranged in the dimensionless space, thereby estimating the distribution of the plotted points in the dimensionless space, that is, the distribution of the state value of the device. As for the method of estimating the distribution of the state value of the device, it is sufficient to estimate the probability density distribution of the plotted points in the dimensionless space, for example, the maximum likelihood method, Bayesian estimation or EM algorithm can be used.

[0048] The output unit 17 outputs data used in monitoring the state of the device. The data used in monitoring the state of the device is, for example, a dimensionless space, which uses a common shape to represent multiple display shapes that respectively represent multiple normal information, and the distribution of the state value of the device is configured for the area represented by the common shape. The output unit 17, for example, causes a display device to display the dimensionless space. In addition, the output unit 17 may also be a structural element of an external device that is provided separately from the device state monitoring device 1. In this case, the device state monitoring device 1 does not have the output unit 17, and the distribution of the state value of the device is displayed on the display device of the external device through the output unit 17 of the external device.

[0049] In addition, the equipment status monitoring device 1 may also have only the correction unit 15 and the output unit 17, and the data acquisition unit 11, the normal information acquisition unit 12, the position relationship acquisition unit 13, the projection unit 14 and the distribution estimation unit 16 are structural elements possessed by an external device provided separately from the equipment status monitoring device 1.

[0050] In this case, in the device state monitoring device 1, the correction unit 15 corrects both or either one of the normal information and the measured data in the dimensionless space based on the relationship between the measured data representing the normal state of the device in the dimensionless space obtained from the projection unit 14 of the external device and the display shape representing the normal information. Then, the output unit 17 causes the display device of the device state monitoring device 1 or the display device of the external device to display the dimensionless space in which the distribution of the device state is projected with respect to the common shape of the plurality of normal information representing the device.

[0051] Figure 2 1 is a flowchart showing the device status monitoring method of embodiment 1, showing the operation of the device status monitoring device 1. The measured data representing the status value of the device is the sensor data shown below. The data acquisition unit 11 acquires the sensor data representing the status value of the monitored device (step ST1). For example, the data acquisition unit 11 acquires the time series of sensor data for a certain period of time (hereinafter referred to as the monitoring period) during which the device is monitored by continuously or periodically inputting the sensor data detected by the sensor installed in the device, that is, the time series data of the status value of the device. In addition, the sensor data can also be stored in a storage device, and the data acquisition unit 11 can acquire the sensor data from the storage device.

[0052] Next, the normal information acquisition unit 12 acquires normal information indicating the normal state of the monitored device (step ST2). For example, when the monitored device is an air conditioner, and the display shape indicating the normal information is the refrigeration cycle graph data on the ph line diagram in the normal operation state of the air conditioner, the normal information acquisition unit 12 uses the sensor data detected from the air conditioner in the normal operation state by the pressure sensor and the temperature sensor provided in the air conditioner to calculate the refrigeration cycle graph data in the normal operation state of the air conditioner. In addition, the refrigeration cycle graph data in the normal operation state of the air conditioner may be stored in a storage device, and the normal information acquisition unit 12 may sequentially acquire the refrigeration cycle graph data corresponding to the operation state during the monitoring period from the storage device.

[0053] The positional relationship acquisition unit 13 acquires the positional relationship of the sensor data in the display shape indicating the normal information of the monitoring target device (step ST3 ). Figure 3A is a graph showing the relationship between multiple display shapes representing multiple normal information of the device and multiple sensor data of the device, and shows the relationship between sensor data (1) and sensor data (2) detected by the sensor installed in the device. For example, the position relationship acquisition unit 13 generates a position relationship using the measured data acquired by the data acquisition unit 11 and the normal information acquired by the normal information acquisition unit 12. Figure 3 A is the graph shown.

[0054] exist Figure 3 The graph shown in A is provided with plot points of a plurality of sensor data A consisting of sensor data (1) and sensor data (2), and a refrigeration cycle graph B showing a plurality of normal information corresponding to these plot points is displayed. Figure 3 As shown in A, even if the operating conditions determined by sensor data (1) and sensor data (2) are the same, the shape of the refrigeration cycle graph B changes in various ways depending on the external environment or operating conditions of the air conditioner. Figure 3 As shown in A, the correspondence between the plotted points of the sensor data A and the refrigeration cycle graph B of the normal information becomes complicated, and the state of the equipment cannot be accurately identified.

[0055] Therefore, the device state monitoring device 1 projects the multiple measured data detected from the device into a dimensionless space, and the dimensionless space represents multiple display shapes that respectively represent multiple normal information with a common shape. Thus, the device state monitoring device 1 can identify multiple states of the device based on the common display shape. As information for projecting the measured data into the dimensionless space, the positional relationship acquisition unit 13 acquires the positional relationship of the measured data in the display shape representing the normal information. The positional relationship is, for example, the distance information between the center point of the display shape representing the normal information and the measured data, and the angle information formed by a straight line passing through the center point of the display shape representing the normal information and the measured data and the side in the display shape representing the normal information that intersects with the straight line.

[0056] The projection unit 14 projects the plurality of sensor data onto a dimensionless space based on the positional relationship between the plurality of display shapes respectively indicating the plurality of normal information of the device and the plurality of sensor data detected from the device (step ST4 ). Figure 3 B is a graph showing a dimensionless space into which multiple sensor data of a device are projected. Figure 3In B, the projection unit 14 applies the center point of the display shape representing the plurality of normal information corresponding to the plurality of sensor data to the center 21 of the circle 20 in the dimensionless space, converts the positional relationship between the center point of the display shape representing the normal information and the sensor data into a positional relationship with the center 21 of the circle 20, thereby arranging the plotted points of the plurality of sensor data on the circle 20. As a result, the distribution 22 of the sensor data is arranged on the circle 20.

[0057] The positions of the plotted points of the sensor data in the dimensionless space are positions corresponding to the relationship between the sensor data before projection and the display shape representing normal information. Figure 3 Component (1) shown in B is a component corresponding to the change of sensor data (1), and component (2) is a component corresponding to the change of sensor data (2).

[0058] The correction unit 15 corrects the normal information and the sensor data based on the relationship between the sensor data and the normal information (step ST5 ). Figure 4 FIG. 2 is an explanatory diagram showing an overview of the correction process of the normal information of the device. Figure 4 In FIG. 1 , the display shape representing the normal information of the device is a circle 20 with a radius of 1 set in the dimensionless space. The center 21 of the circle 20 is the origin of the coordinate system of the dimensionless space. The sensor data detected from the device in the normal operation state is mostly distributed in the center 21 of the circle 20.

[0059] However, if Figure 4 As shown in the left graph in FIG. 1 , in the dimensionless space before the normal information of the device is corrected, the center 23 of the distribution 22 of the sensor data is offset from the center 21 of the circle 20. The offset between the center 21 of the circle 20 in the coordinate system of the dimensionless space and the center 23 of the distribution 22 of the sensor data is equivalent to the error between the normal information in the actual operating environment and the normal information assuming an ideal operating environment.

[0060] The correction unit 15 corrects the normal information in a direction to eliminate the deviation between the center 21 of the circle 20 and the center 23 of the distribution 22 of the sensor data. That is, the normal information of the device is corrected to be represented by the circle 20A centered on the center 23 of the distribution 22 of the sensor data. For example, the correction unit 15 searches for the closest point 24 between the center 23 of the distribution 22 of the sensor data and the circle 20. The position of the center 23 of the distribution 22 of the sensor data is set to, for example, a position corresponding to the average value or the central value of the coordinate values ​​of the plotted points of the sensor data arranged in the dimensionless space in the X-axis direction and the Y-axis direction.

[0061] The correction unit 15 calculates the distance r' between the searched nearest point 24 and the center 23 of the distribution 22 of the sensor data, and corrects the normal information so that Figure 4 As shown in the right graph in FIG. 1 , the circle 20A with a radius of r' is used to represent the distribution 22 of the sensor data. Then, the correction unit 15 may correct the deviation of the distribution 22 of the sensor data to 1 / r' times. For example, the correction unit 15 arranges a new distribution 22A in which the distance between the center 23 of the distribution 22 and each plotted point is 1 / r' times on the circle 20A. In this way, in addition to the normal information of the device, the measured data is corrected.

[0062] The distribution estimation unit 16 estimates the distribution of the state of the device based on the plurality of sensor data projected into the dimensionless space (step ST6). For example, the distribution estimation unit 16 performs a mixed Gaussian model estimation process on the plurality of plotted points arranged in the dimensionless space, thereby estimating Figure 4 Distribution 22A in the dimensionless space shown. The color or depth of distribution 22A is determined by the number of plotted points corresponding to the sensor data, and the number of plotted points corresponding to the sensor data is the largest in the darkest area and decreases as the color becomes lighter. The output unit 17 displays the dimensionless space in which distribution 22A is configured on the display device.

[0063] Figure 5 is a flowchart showing the positional relationship calculation process, showing Figure 2 The positional relationship acquisition unit 13 displays a two-dimensional coordinate plane on which the plot points of the sensor data of the device acquired by the data acquisition unit 11 are arranged and the shapes representing the normal information of the device acquired by the normal information acquisition unit 12 are arranged (step ST1a).

[0064] Figure 6 is a graph showing the relationship between the display shape representing normal information of a device and the sensor data detected from the device. Figure 6 In FIG. 1 , sensor data 31 is data indicating a state value of a device determined based on sensor data (1) and sensor data (2) detected from the monitored device. Display shape 32 is normal information related to the operating state of the device indicated by sensor data 31 and is a quadrilateral having vertices 33 to 36.

[0065] The positional relationship acquisition unit 13 calculates the center 37 of the display shape 32 (step ST2 a ). Figure 7 3 is a graph showing a display shape 32 indicating normal information of a device, sensor data 31 detected from the device, and a center 37. For example, the positional relationship acquisition unit 13 calculates the center of gravity of the outer shape of the display shape 32 as the center 37. Figure 7 In FIG. 3 , the display shape 32 is a quadrilateral, and therefore the positional relationship acquisition unit 13 calculates the intersection of the diagonal line connecting the vertex 33 and the vertex 35 and the diagonal line connecting the vertex 34 and the vertex 36 as the center 37 .

[0066] Figure 8 3 is a graph showing the positional relationship between the display shape 32 representing normal information of a device and the sensor data 31 detected from the device. Figure 8 As shown, the positional relationship acquisition unit 13 calculates the distance r between the sensor data 31 and the center 37 of the display shape 32. a In addition, the angle θ between the straight line A passing through the sensor data 31 and the center 37 of the display shape 32 and the side C of the display shape 32 intersecting the straight line A is calculated. a (Step ST3a) Next, the positional relationship acquisition unit 13 calculates the distance r from the center 37 of the display shape 32 to the intersection point Ca of the straight line A and the side C of the display shape 32. n .

[0067] The positional relationship acquisition unit 13 calculates the distance r a The length relative to the straight line A is the distance r n The relative ratio r R (=r a / r n ). Next, the positional relationship acquisition unit 13 calculates the angle θ between the straight line A and the side C. a The relative ratio r calculated in this way R and angle θ a The data representing the positional relationship of the sensor data 31 with respect to the center 37 of the display shape 32 is output to the projection unit 14 .

[0068] Fig. 9 is a flowchart showing the projection process of sensor data onto dimensionless space, showing Figure 2 The details of the processing of step ST4 are shown in FIG. Fig.10 is a graph showing an example of a dimensionless space. A dimensionless space can be Fig.10 The space shown is set with a circle 20 of radius 1. When the angle from the axis of component (1) is θ, the point on circle 20 is represented by (1,θ). Component (1) is Figure 6 The component corresponding to the change of sensor data (1) and component (2) are Figure 6 The component corresponding to the change of sensor data (2).

[0069] The projection unit 14 projects the sensor data (measured data) 31 onto the dimensionless space based on the positional relationship of the sensor data 31 in the display shape 32 (step ST1 b ). Fig.11 is shown projected onto Fig.10 For example, the projection unit 14 will Figure 8 The center 37 of the display shape 32 is shown to apply to Fig.10The center 21 of the circle 20 in the dimensionless space shown. Next, the projection unit 14 replaces the distance r by n The relative ratio r of the distance from the center 37 of the display shape 32 to the sensor data 31 is used. R and the angle θ between the straight line A passing through the sensor data 31 and the side C of the display shape 32 a , thereby arranging the plotted points of the sensor data 25 in a dimensionless space. Fig.11 As shown, the plotted points of the sensor data 25 are converted into points (r R ,θ a ) and are arranged in dimensionless space.

[0070] The projection unit 14 checks whether all the sensor data 31 that have acquired the positional relationship with the display shape 32 are projected into the dimensionless space (step ST2b). Here, if there is sensor data 31 that has not been projected into the dimensionless space (step ST2b: No), the process returns to step ST1b, and the centers 37 of the plurality of display shapes 32 are sequentially applied to the center 21 of the circle 20, and the sensor data 31 are projected into the dimensionless space. In the case that there is no sensor data 31 that has not been projected into the dimensionless space (step ST2b: Yes), Fig. 9 The processing is finished.

[0071] The correction unit 15 corrects both or either one of the normal information and the sensor data in the dimensionless space based on the relationship between the sensor data indicating the normal state of the device in the dimensionless space and the display shape indicating the normal information. Fig.12 : is a graph showing an overview of the correction process of the sensor data of the device. Assuming that the device is not degraded, the distribution of the sensor data A obtained from the device in a normal operating state is distributed near the center 37 of the normal information. However, when there is an individual difference in the device or the information required for calculating the normal information is insufficient, the normal information of the device will have errors. Therefore, Fig.12 As shown, an offset 40 occurs between the sensor data A and the center 37 of the displayed shape B.

[0072] like Fig.12 As shown, when similar display shapes B are extracted from a plurality of display shapes B (e.g., a refrigeration cycle diagram) respectively representing a plurality of normal information, the devices having similar display shapes B are in similar operating states. There is a tendency that a plurality of sensor data A representing similar operating states are distributed at close positions in the display shape B representing the normal information of the device. The correction unit 15 estimates the offset 40 between the sensor data A and the center 37 of the display shape B using this tendency, thereby correcting the sensor data A and the display shape B.

[0073] Fig.13 is a flowchart showing the projection processing and correction processing of sensor data of the device, showing Figure 2 The details of the processing of step ST4 and step ST5 in . Fig.14 is a graph showing the positional relationship of sensor data (hereinafter referred to as measured data) of the device in a dimensionless space. Fig.14 In the dimensionless space shown, a circle 20 with a radius of 1 representing normal information of the device is arranged, and plot points of the measured data are arranged (step ST1c).

[0074] The calibration unit 15 calculates the position (x, y) of the plotted point of the measured data of the device in the dimensionless space with the center 21 of the circle 20 as the origin (step ST2c). In addition, the position (x, y) of the plotted point of the measured data of the device is a position indicating an offset from the origin of the dimensionless space. The position in the dimensionless space can also be expressed in polar coordinate form.

[0075] Next, the calibration unit 15 determines whether the measured data of the calculated position (x, y) is data obtained within the initial learning period (step ST3c). The initial learning period is a period for learning the relationship between the position coordinates of the display shape representing normal information in the dimensionless space and the position coordinates of the plotted points of the measured data obtained from the device in the normal state. The initial learning period is set in advance by the user, for example.

[0076] In the case where the measured data is data obtained during the initial learning period (step ST3c: Yes), the correction unit 15 learns the offset from the origin of the measured data based on the relationship between the normal information and the measured data (step ST4c). For example, the correction unit 15 obtains a learner that learns the correlation between the position (x, y) of the plotted points of the measured data detected from the device in the normal state during the initial learning period and the position coordinates of the display shape representing the normal information of the device in the dimensionless space. In this case, the learning data is the position of the plotted points of the measured data detected from the device in the normal state and the position of the display shape representing the normal information of the device in the dimensionless space.

[0077] The learner takes the position coordinates of the display shape representing the normal information of the device in the dimensionless space as input, and estimates the position (x', y') of the estimated data representing the normal state value of the device in various operating states in the dimensionless space. The position (x', y') of the estimated data is a position representing the offset from the origin of the dimensionless space. In addition, the position coordinates of the display shape representing the normal information of the device in the dimensionless space are, for example, Figure 6The coordinates of the positions of the vertices 33 to 36 of the quadrilateral representing the normal information of the device are shown. The learner may also use multiple regression analysis or a neural network for estimation.

[0078] When the learner is acquired during the initial learning period or the measured data is not data obtained during the initial learning period but the learner is acquired after the initial learning period (step ST3c: No), the correction unit 15 estimates the position (x', y') of the estimated data by using the learner (step ST5c). Figure 6 The position coordinates of the vertices 33 to 36 of the quadrilateral representing the normal information of the device are input to the learner, thereby estimating the position (x', y') of the estimated data in the dimensionless space.

[0079] The correction unit 15 corrects the position (x, y) of the plotted point of the measured data detected after the initial learning period using the position (x', y') of the estimated data corresponding to the state represented by the measured data (step ST6c). For example, the correction unit 15 subtracts the position (x', y') of the estimated data from the position (x, y) of the plotted point of the measured data of the device configured in the dimensionless space, and corrects the measured data so that the offset from the origin of the dimensionless space becomes smaller. Thus, the position (x-x', y-y') of the corrected measured data is obtained. In this way, the measured data detected from the device after the initial learning period is corrected.

[0080] In addition, the learning data of the learner may also be measured data detected from a device in a normal state. That is, the learner uses the measured data detected from a device in a normal state as learning data, and learns the correlation between the measured data detected from the device in a normal state and the position of the display shape representing the normal information of the device in the dimensionless space. In this way, a learner is obtained that takes the position coordinates of the display shape representing the normal information of the device in the dimensionless space as input, and estimates the position (x', y') of the estimated data representing the normal state value of the device in various operating states in the dimensionless space.

[0081] The distribution estimation unit 16 performs a mixed Gaussian model estimation process on the plotted points of the plurality of measured data whose positions in the dimensionless space have been corrected by the correction unit 15, thereby estimating, for example, Figure 3 B shows the distribution of measured data in dimensionless space 22. Fig.15 A is a graph showing the distribution 22 of the sensor data projected into the dimensionless space. In the distribution 22 of the sensor data, the number of plotted points of the sensor data is greatest in the darkest region 28 and decreases as the color becomes lighter.

[0082] exist Fig.15In A, component (1) is a component corresponding to the change of sensor data (1), and component (2) is a component corresponding to the change of sensor data (2). Regarding region 28, the number of plotted points of sensor data is the largest in distribution 22, and the sensor data in region 28 has a dominant influence on the characteristics of distribution 22. Circle 20 is a display shape indicating the normal state of the device represented by sensor data composed of sensor data (1) and sensor data (2).

[0083] The distribution estimation unit 16 determines whether the state of the monitored device is close to the normal state based on the positional relationship between the area 28 of the distribution 22 and the circle 20. The area 28 of the distribution 22 exists at a position separated from the center 21 of the circle 20, so it can be known that the state of the device deviates from the normal state, that is, the state of the device is deteriorating. In this way, the device state monitoring device 1 can determine the change trend of the state of the device during the monitoring period, and even if an accidental outlier occurs in the sensor data, its influence is reduced.

[0084] Fig.15 B is an explanatory diagram showing an overview of the process of estimating the state of the device based on the distribution of sensor data projected into the dimensionless space. Fig.15 In B, distribution 22A is the distribution of the state of the device estimated by the distribution estimation unit 16, and is estimated based on the sensor data obtained during the monitoring period (1). Distribution 22B is the distribution of the same state as distribution 22A, but is estimated based on the sensor data obtained during the monitoring period (2) which is longer than the monitoring period (1). In addition, the outer shapes of distributions 22A and 22B are elliptical.

[0085] The state of the monitored device changes over time, and the distribution 22A shifts to the distribution 22B. At this time, the shift from the center point 23A of the distribution 22A to the center point 23B of the distribution 22B can be used (r c ,θ c ) indicates the distance r c is the distance between the center point 23A and the center point 23B, and the angle θ c is the angle formed by the line segment connecting the center point 23A and the center point 23B at the center point 23A. Similarly, the transition from the point 27A on the major axis of the distribution 22A to the point 27B on the major axis of the distribution 22B can be calculated using (r m ,θ m ) indicates that the transition from point 28A on the minor axis of distribution 22A to point 28B on the minor axis of distribution 22B can be expressed by (r s ,θ s )express.

[0086] The distribution estimation unit 16 determines whether the distribution of the sensor data projected into the dimensionless space is different from, for example, (r, θ) based on (r, θ) indicating the transition of the distribution of the measured data projected into the dimensionless space. Fig.15 The positional relationship between the circles 20 shown in A is determined, and whether the state of the monitored device has deteriorated is determined based on the determined positional relationship. Thus, by quantifying the temporal change of the distribution estimated by the distribution estimation unit 16, it is possible to accurately determine whether the state of the monitored device has deteriorated.

[0087] In the device state monitoring apparatus 1 , the display shape of the normal information representing the device in the dimensionless space may be a circle or a polygon.

[0088] The functions of the data acquisition unit 11, the normal information acquisition unit 12, the positional relationship acquisition unit 13, the projection unit 14, the correction unit 15, the distribution estimation unit 16 and the output unit 17 in the device state monitoring device 1 are realized by the processing circuit. Figure 2 The processing circuit may be dedicated hardware, or may be a CPU (Central Processing Unit) that executes a program stored in a memory.

[0089] Fig.16 A is a block diagram showing a hardware configuration for realizing the functions of the equipment state monitoring device 1 . Fig.16 B is a block diagram showing a hardware structure for executing software that realizes the functions of the equipment status monitoring device 1. Fig.16 A and Fig.16 In B, the input interface 100 is an interface for relaying sensor data output from a sensor installed in the monitoring target device to the device state monitoring apparatus 1. In addition, the output interface 101 is an interface for relaying information output from the device state monitoring apparatus 1.

[0090] The processing circuit is Fig.16In the case of the dedicated hardware processing circuit 102 shown in A, the processing circuit 102 is, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a component obtained by combining them. The functions of the data acquisition unit 11, the normal information acquisition unit 12, the position relationship acquisition unit 13, the projection unit 14, the correction unit 15, the distribution estimation unit 16, and the output unit 17 in the equipment status monitoring device 1 can be realized by different processing circuits, or their functions can be realized uniformly by one processing circuit.

[0091] The processing circuit is Fig.16 In the case of the processor 103 shown in B, the functions of the data acquisition unit 11, the normal information acquisition unit 12, the positional relationship acquisition unit 13, the projection unit 14, the correction unit 15, the distribution estimation unit 16 and the output unit 17 in the device state monitoring device 1 are implemented by software, firmware or a combination of software and firmware. In addition, the software or firmware is described as a program and stored in the memory 104.

[0092] The processor 103 reads and executes the program stored in the memory 104, thereby realizing the functions of the data acquisition unit 11, the normal information acquisition unit 12, the position relationship acquisition unit 13, the projection unit 14, the correction unit 15, the distribution estimation unit 16 and the output unit 17 in the device state monitoring device 1. For example, the device state monitoring device 1 has the memory 104, which is used to store the program that is executed by the processor 103. Figure 2 The memory 104 may be a computer-readable storage medium storing a program for causing a computer to function as the data acquisition unit 11, the normal information acquisition unit 12, the positional relationship acquisition unit 13, the projection unit 14, the correction unit 15, the distribution estimation unit 16, and the output unit 17.

[0093] The memory 104 is, for example, a nonvolatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), EEPROM (Electrically-EPROM), a magnetic disk, a floppy disk, an optical disk, a high-density disk, a mini disk, a DVD, etc.

[0094] It is also possible to implement a part of the functions of the data acquisition unit 11, the normal information acquisition unit 12, the positional relationship acquisition unit 13, the projection unit 14, the correction unit 15, the distribution estimation unit 16, and the output unit 17 in the device state monitoring device 1 by using dedicated hardware, and implement a part by using software or firmware. For example, the functions of the data acquisition unit 11, the normal information acquisition unit 12, and the positional relationship acquisition unit 13 are implemented by the processing circuit 102 which is dedicated hardware, and the functions of the projection unit 14, the correction unit 15, the distribution estimation unit 16, and the output unit 17 are implemented by the processor 103 reading and executing the program stored in the memory 104. In this way, the processing circuit can implement the above functions by hardware, software, firmware, or a combination thereof.

[0095] As described above, the device state monitoring device 1 of embodiment 1 projects a plurality of measured data representing the state value of the device into a dimensionless space, estimates the distribution of the state of the device based on the plurality of measured data projected into the dimensionless space, and the dimensionless space represents a plurality of display shapes with a common shape, and the plurality of display shapes respectively show a plurality of normal information representing the normal state of the device. Furthermore, based on the relationship between the measured data representing the normal state of the device in the dimensionless space and the display shape representing the normal information, both or either of the normal information and the measured data in the dimensionless space are corrected. Thus, the device state monitoring device 1 can accurately identify the state of the device.

[0096] In the device state monitoring device 1 of the first embodiment, the positional relationship acquisition unit 13 calculates the distance between the center point of the display shape representing the normal information and the measured data, and calculates the angle between the straight line passing through the center point and the measured data and the side of the display shape representing the normal information intersecting with the straight line. Thus, the information showing the positional relationship of the measured data in the display shape representing the normal information can be accurately acquired.

[0097] In the device state monitoring apparatus 1 of the first embodiment, the correction unit 15 includes a learner, which takes the position of the display shape representing the normal information in the dimensionless space as an input, estimates the position of the data representing the normal state value of the device in the dimensionless space, and corrects the measured data using the position of the data estimated by the learner so that the offset from the origin of the dimensionless space becomes smaller. Thus, the offset from the origin of the dimensionless space in the measured data projected into the dimensionless space can be accurately corrected.

[0098] Furthermore, any structural elements of the embodiments may be modified or omitted.

[0099] Industrial Applicability

[0100] The equipment state monitoring device of the present invention can be used to monitor the operating state of an air conditioner, for example.

[0101] Description of symbols

[0102] 1: Equipment status monitoring device; 11: Data acquisition unit; 12: Normal information acquisition unit; 13: Position relationship acquisition unit; 14: Projection unit; 15: Correction unit; 16: Distribution estimation unit; 17: Output unit; 20, 20A: Circle; 22, 22A, 22B: Distribution; 23, 37: Center; 23A, 23B: Center point; 24: Nearest point; 25, 31: Sensor data; 28: Area; 32: Display shape; 33-36: Vertex; 40: Offset; 100: Input interface; 101: Output interface; 102: Processing circuit; 103: Processor; 104: Memory.

Claims

1. A device for monitoring equipment status, It is characterized in that The equipment status monitoring device has: a data acquisition unit that acquires actual measurement data representing a state value of the device; a normal information acquisition unit that acquires normal information indicating a normal state of the device; a positional relationship acquisition unit that acquires a positional relationship of the actual measurement data in a display shape representing the normal information of the device; a projection unit, which projects the plurality of measured data onto a dimensionless space according to a positional relationship between a plurality of display shapes respectively representing a plurality of the normal information of the device and the plurality of measured data, wherein the dimensionless space represents the plurality of display shapes respectively representing the plurality of the normal information with a common shape; a correction unit that corrects both or either one of the normal information and the measured data in the dimensionless space based on a relationship between the measured data representing a normal state of the device in the dimensionless space and a display shape representing the normal information; and A distribution estimating unit estimates a distribution of the state of the device based on the plurality of measured data projected into the dimensionless space.

2. The device state monitoring device according to claim 1, It is characterized in that As information showing the positional relationship of the measured data in the display shape representing the normal information, the positional relationship acquisition unit calculates the distance between the center point of the display shape representing the normal information and the measured data, and calculates the angle formed by a straight line passing through the center point and the measured data and an edge of the display shape representing the normal information intersecting the straight line.

3. The device state monitoring device according to claim 1, It is characterized in that The correction unit has a learner that takes the position of the display shape representing the normal information in the dimensionless space as input, estimates the position of the data representing the normal state value of the device in the dimensionless space, and uses the position of the data estimated by the learner to correct the measured data so that the offset from the origin of the dimensionless space becomes smaller.

4. A device status monitoring device, the device status monitoring device monitors the status of a device, It is characterized in that The equipment status monitoring device has: an output unit that outputs data used in monitoring the state of the device; and a correction unit that corrects the measured data indicating the state value of the device or the normal information indicating the normal state of the device, The output unit outputs a dimensionless space, wherein the dimensionless space represents a plurality of display shapes respectively representing a plurality of the normal information of the device with a common shape, and a distribution of the state of the device is projected onto the common shape, The correction unit corrects both or either one of the normal information and the measured data in the dimensionless space based on a relationship between the measured data indicating a normal state of the device in the dimensionless space and a display shape indicating the normal information.

5. The device state monitoring device according to claim 4, It is characterized in that The device state monitoring device comprises: a data acquisition unit that acquires the measured data; a normal information acquisition unit, which acquires the normal information; a positional relationship acquisition unit that acquires a positional relationship of the actual measurement data in a display shape representing the normal information; a projection unit configured to project the plurality of measured data onto the dimensionless space according to a positional relationship between a plurality of display shapes respectively representing a plurality of the normal information of the device and the plurality of measured data; as well as A distribution estimating unit estimates a distribution of the state of the device based on the plurality of measured data projected into the dimensionless space.

6. The device state monitoring device according to claim 5, It is characterized in that As information showing the positional relationship of the measured data in the display shape representing the normal information, the positional relationship acquisition unit calculates the distance between the center point of the display shape representing the normal information and the measured data, and calculates the angle formed by a straight line passing through the center point and the measured data and an edge of the display shape representing the normal information intersecting the straight line.

7. The equipment status monitoring device according to claim 4 or 5, It is characterized in that The correction unit has a learner that takes the position of the display shape representing the normal information in the dimensionless space as input, estimates the position of the data representing the normal state value of the device in the dimensionless space, and uses the position of the data estimated by the learner to correct the measured data so that the offset from the origin of the dimensionless space becomes smaller.

8. A method for monitoring device status, It is characterized in that The device status monitoring method has the following steps: The data acquisition unit acquires actual measurement data representing a state value of the device; The normal information acquisition unit acquires normal information indicating a normal state of the device; A positional relationship acquisition unit acquires a positional relationship of the actual measurement data in a display shape representing the normal information of the device; The projection unit projects the plurality of measured data onto a dimensionless space according to a positional relationship between a plurality of display shapes respectively representing a plurality of the normal information of the device and the plurality of measured data, wherein the dimensionless space represents the plurality of display shapes respectively representing the plurality of the normal information with a common shape; The correction unit corrects both or either one of the normal information and the measured data in the dimensionless space according to a relationship between the measured data representing a normal state of the device in the dimensionless space and a display shape representing the normal information; and The distribution estimating unit estimates the distribution of the state of the device based on the plurality of measured data projected into the dimensionless space.

Images