JP6749488B2 - Abnormality importance calculation system, abnormality importance calculation device, and abnormality importance calculation program - Google Patents

Description

The present invention relates to an abnormality importance degree calculation system, an abnormality importance degree calculation device, and an abnormality importance degree calculation program.

2. Description of the Related Art Conventionally, in order to detect an abnormality of equipment including equipment installed in a building or plant equipment, a sensor group is provided in the equipment. If any of the sensors in the sensor group detects an abnormal value, it is suspected that the device has some trouble.

The simplest method for determining whether or not the detected value of the sensor is an abnormal value is to set a threshold value for the sensor in advance and make a determination based on a comparison between the detected value of the sensor and the threshold value. However, even when the device is operating normally, the detected value of the sensor is not always constant, and it may fluctuate depending on the operating conditions, operating environment, or deterioration over time. Setting a threshold is not easy.

In view of this, conventionally, a prediction model is constructed based on past detection values of the sensor detected when the device is normally operating, and based on the prediction model, detection of the sensor during normal operation of the device Values are being predicted. Then, it is determined whether or not the detected value of the sensor is an abnormal value based on the comparison between the detected predicted value and the detected value (generally, the difference between the detected predicted value and the detected value).

For example, in Patent Document 1, a multi-body correlation model showing the correlation between the detection values of each sensor is acquired based on the past detection values of a plurality of sensors provided in the device, and the detection values of each sensor are It is disclosed that when applied to a body correlation model, an abnormality in the detection value of each sensor is detected based on whether model destruction has occurred in the multibody correlation model.

Further, in Patent Document 2, a model is constructed by learning past detection values of a plurality of sensors when the device is normally operating, and a plurality of sensor detection values are applied to the model, It is disclosed to detect an anomaly with respect to a detection value of the sensor.

Further, in Patent Document 3, a normal operation pattern database is constructed based on past operation data when the device is normally operating, and the correlation between the operation data detected by the detection means and the normal operation pattern database is provided. It is disclosed that a relationship value is obtained and an abnormality is detected when the correlation value is less than a threshold value.

International Publication 2016/088362 Pamphlet JP, 2014-56598, A JP, 2011-209847, A

As in the past, when determining whether or not the detection value of the sensor is an abnormal value based on the detection prediction value based on the prediction model and the actual detection value of the sensor, there is no malfunction in the device. Even in this case, it may be determined that the sensor has detected an abnormal value. For example, if the sensor is a temperature sensor, and if the temperature on one day is extremely hot compared to the average year, the detected temperature value of the temperature sensor will increase even if the device is operating normally, causing abnormalities. May detect a value. In such a case, for example, a person who maintains the device (for example, a service person) has to perform useless confirmation work on the originally normal device.

Further, when a device is provided with a sensor group, a plurality of sensors in the sensor group may detect an abnormal value at the same timing. In this case, a person who maintains the device may not be able to know which sensor, out of the plurality of sensors that have detected an abnormal value, should be prioritized for confirmation. As a result, for example, there is a possibility that damage to the device may occur due to the fact that confirmation of a location with a high degree of urgency is delayed.

Regarding the detection value (abnormal value) of the sensor, the abnormality importance, which is an index that indicates whether it is due to a true malfunction in the device or which sensor should be prioritized for checking, is calculated. Thus, the above problem can be solved. That is, a person who maintains the device can determine that, for example, if the abnormality importance level with respect to the sensor detection value is low, it is not due to a true malfunction of the device, and the abnormality importance levels are calculated for a plurality of sensors. By doing so, it is possible to prioritize and confirm the location corresponding to the sensor with high abnormality importance.

An object of the present invention is to calculate the degree of abnormality importance with respect to the detection value of the sensor provided in the device.

The present invention is an abnormality importance degree calculation system including a sensor group provided in a device and a server that receives a detection value detected by the sensor group, wherein the server includes each sensor included in the sensor group. Regarding, based on the difference between the predicted detection predicted value and the actual detection value, an abnormality degree calculation unit that calculates an abnormality degree for the detection value, and the abnormality degree for the detection value of the sensor of interest in the sensor group, Other than the attention sensor in the sensor group, and the abnormality degree with respect to the detection value of the attention sensor at the time of abnormality detection when the abnormality sensor is equal to or more than a predetermined abnormality degree threshold for the attention sensor And an abnormality importance degree calculation unit that calculates an abnormality importance degree with respect to the detection value of the sensor of interest based on the abnormality degree with respect to the detection value of the sensor.

Desirably, the abnormality importance degree calculation unit increases the abnormality importance degree with respect to the detection value of the target sensor as the abnormality degree with respect to the detection value of the other sensor at the time of abnormality detection is smaller, and when the abnormality is detected. The abnormality degree with respect to the detection value of the target sensor is decreased as the degree of abnormality with respect to the detection value of the other sensor increases.

Desirably, at the time of the abnormality detection, the abnormality degree with respect to the detection value of the attention sensor, from the time when it becomes equal to or more than a predetermined abnormality degree threshold value for the attention sensor, to the time when it becomes less than the abnormality degree threshold value. The abnormality importance degree calculation unit has a period, the representative value of the abnormality degree with respect to the detection value of the sensor of interest at a plurality of time points during the abnormality detection, and the abnormality degree with respect to the detection value of the other sensor. The abnormality importance level with respect to the detection value of the sensor of interest is calculated based on the representative value of.

Preferably, each sensor included in the sensor group is provided at a different point in the building.

Desirably, the server notifies a person who maintains the device of abnormality information indicating that the detection value of the attention sensor has become an abnormal value and abnormality information indicating the degree of abnormality importance with respect to the detection value of the attention sensor. And are further included.

Further, the present invention, for each sensor included in the sensor group provided in the device, based on the difference between the predicted detection predicted value and the actual detection value, the abnormality degree calculation unit that calculates the abnormality degree for the detection value. And the abnormality with respect to the detection value of the sensor of interest in the sensor group, the abnormality with respect to the detection value of the attention sensor at the time of abnormality detection when the abnormality level threshold value is predetermined for the attention sensor And an abnormality importance degree calculation unit that calculates an abnormality importance degree with respect to a detection value of the attention sensor based on the abnormality degree with respect to a detection value of a sensor other than the attention sensor in the sensor group. And an abnormality importance degree calculating device.

In addition, the present invention provides a computer, for each sensor included in a sensor group provided in the device, based on the difference between the predicted detection predicted value and the actual detection value, the abnormality level for the detection value is calculated. Degree calculating unit, the detection value of the attention sensor at the time of abnormality detection when the abnormality degree with respect to the detection value of the attention sensor of the sensor group is equal to or more than a predetermined abnormality degree threshold value for the attention sensor And an abnormality importance degree for calculating an abnormality importance degree with respect to a detection value of the attention sensor based on the abnormality degree with respect to a detection value of another sensor that is a sensor other than the attention sensor in the sensor group. An abnormality importance degree calculation program characterized by causing it to function as a calculation unit.

According to the present invention, it is possible to calculate the degree of abnormality importance with respect to the detection value of the sensor provided in the device.

It is a schematic configuration diagram of an abnormality importance degree calculation system according to the present embodiment. It is a schematic block diagram of the server which concerns on this embodiment. It is a graph which shows the detection change value of a sensor, and the time change of a detection value. It is a graph which shows the time change of the abnormality degree of a sensor. 6 is a graph showing an example of a change in an abnormality degree of the sensor A with time. 9 is a graph showing an example of a change in the degree of abnormality of the sensor B with time. 7 is a graph showing an example of a change over time in the degree of abnormality of the sensor C. It is a graph which shows the time change of the abnormality degree of the same sensor A as FIG. 5A. 9 is another graph showing another example of the change over time of the abnormality degree of the sensor B. 7 is another graph showing another example of the change over time of the abnormality degree of the sensor C. It is a figure which shows the example of an abnormality information display screen. It is a flow chart which shows a flow of processing of a server concerning this embodiment.

Hereinafter, embodiments of the present invention will be described.

FIG. 1 shows a schematic configuration diagram of an abnormality importance degree calculation system 10 according to the present embodiment. The abnormality importance degree calculation system 10 includes a sensor group 12 including a plurality of sensors, a server 14 as an abnormality importance degree calculation device, and a serviceman terminal 16.

In the present embodiment, the sensor group 12 is provided in each device (facility) in the building 18. The sensors included in the sensor group 12 are sensors that detect various parameters related to the equipment in the building 18, particularly parameters for determining whether or not the equipment is operating normally. Each sensor included in the sensor group 12 may be various types of sensors such as a temperature sensor, a humidity sensor, a voltage sensor, a current sensor, a pressure sensor, a rotation speed sensor, and an acceleration sensor.

Each sensor is provided at a different point in the building 18. In the present specification, the “point” means a certain range including a place (for example, a floor or a room in the case of the building 18) or a system (for example, a duct system or a piping system in the case of the building 18). Is. In this specification, as shown in FIG. 1, an example will be described in which the sensor A is provided on the first floor of the building 18, the sensor B is provided on the second floor of the building 18, and the sensor C is provided on the third floor of the building 18. Of course, the sensor group 12 may be provided at each of the other points. Also, a plurality of sensors may be provided at one point.

Each sensor included in the sensor group 12 is connected to the central monitoring device 20 of the building 18 by wire or wirelessly, and the detection value of each sensor is sequentially sent to the central monitoring device 20. Each sensor can acquire the time when the detection value is detected, and each sensor associates the detection value, the detection time information indicating the detection time, and the information (sensor ID) for identifying the own sensor with the central monitoring device. Send to 20. Hereinafter, the information in which the detected value, the detected time information, and the sensor ID are associated with each other will be referred to as detected value information. In this way, the central monitoring device 20 collects the detection value information from the sensor group 12 provided in the building 18. The detection time may be acquired by the central monitoring device 20. That is, the time when the central monitoring device 20 receives the detection value from each sensor may be acquired as the detection time.

The central monitoring device 20 is communicatively connected to the server 14 via a communication line 22 including the Internet or a LAN (Local Area Network). Each detection value information from the sensor group 12 collected in the central monitoring device 20 is sent to the server 14.

The central monitoring device 20 may be a computer, and at least a receiving unit that receives each detection value information from the sensor group 12, a storage unit that stores each detection value information, and transmits each detection value information to the server 14. It suffices to include a transmitting unit that controls the above and a control unit that controls these.

The server 14 is a computer having a performance that allows it to function as a server. The server 14 is communicatively connected to the central monitoring device 20 and the serviceman terminal 16 via a communication line 22. Details of the server 14 will be described later.

The serviceman terminal 16 is a terminal used by a serviceman who is a person who maintains equipment in the building 18. In the present embodiment, the serviceman terminal 16 is a mobile terminal carried by a serviceman, such as a smartphone or a tablet terminal, but is not limited to this, and may be a stationary computer. The serviceman terminal 16 includes a communication unit including a network adapter, a display unit including a liquid crystal panel, an input unit including a touch panel or buttons, a storage unit including a memory, and a control unit including a microcomputer. Composed. The service person can access the server 14 using the service person terminal 16 and acquire various information. Alternatively, the serviceman terminal 16 can receive the information pushed from the server 14.

FIG. 2 shows a schematic configuration diagram of the server 14.

The communication unit 30 is composed of, for example, a network adapter. The communication unit 30 is for communicating with the central monitoring device 20 and the serviceman terminal 16. Specifically, the communication unit 30 has a function of receiving each detection value information regarding the sensor group 12 from the central monitoring device 20, a function of receiving a connection request from the serviceman terminal 16, or various information. It has a function to send to.

The storage unit 32 includes, for example, a hard disk, a ROM (Read Only Memory), a RAM (Random Access Memory), or the like. The storage unit 32 stores an abnormality importance degree calculation program for operating each unit of the server 14. Alternatively, the storage unit 32 stores various control data and various processing data. Further, as shown in FIG. 2, a past detection value information DB (database) 34 is defined and a prediction model 36 is stored in the storage unit 32.

In the past detected value information DB 34, a past detected value information group regarding the sensor group 12 is accumulated and stored. Specifically, every time each detection value information regarding the sensor group 12 is received from the central monitoring device 20, the control unit 38 described later stores the received detection value information in the past detection value information DB 34. As a result, the past detected value information group regarding the sensor group 12 is accumulated and stored in the past detected value information DB 34.

The prediction model 36 is a model for calculating a detection prediction value predicted to be detected by each sensor included in the sensor group 12 if the device is operating normally. The prediction model 36 is constructed by a prediction value calculation unit 40, which will be described later, based on the past detection value information group stored in the past detection value information DB 34. In this embodiment, a plurality of prediction models 36 corresponding to each sensor are constructed. Details of the prediction model 36 will be described later together with the prediction value calculation unit 40.

The control unit 38 includes, for example, a CPU (Central Processing Unit) and the like. The control unit 38 controls each unit of the server 14 according to the abnormality importance degree calculation program stored in the storage unit 32. Further, as shown in FIG. 2, the control unit 38 also functions as a predicted value calculation unit 40, an abnormality degree calculation unit 42, an abnormality determination unit 44, an abnormality importance degree calculation unit 46, and an abnormality notification unit 48. Hereinafter, these functions performed by the control unit 38 will be described.

The predicted value calculation unit 40 includes a plurality of pieces of detection value information (hereinafter referred to as detection value information acquired by the sensor group 12 during the normal operation of each device in the building 18 among the detection value information groups stored in the past detection value information DB 34. Based on “normal detection value information”), a detection prediction value predicted to be detected by each sensor included in the sensor group 12 if the device is operating normally is calculated. As a method of identifying the normal detection value information from the detection value information group stored in the past detection value information DB 34, for example, when the control unit 38 stores the detection value information in the past detection value information DB 34, an abnormality is detected. If not, the detection value information is flagged and stored, and the predicted value calculation unit 40 identifies the normal detection value information from the detection value information group based on the presence or absence of the flag. You may make it extract by extracting. Alternatively, the control unit 38 may not store the detection value information when the abnormality is detected in the past detection value information DB 34 from the beginning. In this way, all the detection value information groups stored in the past detection value information DB 34 become normal detection value information.

First, the predicted value calculation unit 40 builds the predicted model 36 based on the normal time detected value information. As a method for constructing the prediction model 36, a known machine learning method can be used. For example, based on the detection time and the detection value included in the normal-time detection value information, by using a time-series analysis method of predicting the detection value of the sensor by performing a time-series analysis of the time change of the past detection value of the sensor. The prediction model 36 can be constructed by

The prediction value calculation unit 40 builds a prediction model 36 for each sensor included in the sensor group 12. Specifically, the predicted value calculation unit 40 specifies the normal detection value information for the sensor to be processed based on the sensor ID included in the normal detection value information, and based on the specified normal detection value information. Then, the prediction model 36 for the sensor is constructed. As a result, the storage unit 32 stores a plurality of prediction models 36 corresponding to the respective sensors.

Next, the prediction value calculation unit 40 calculates the detection prediction value predicted to be taken by the detection value of each sensor based on the plurality of prediction models 36 constructed for each sensor if the device is operating normally. calculate. For example, the prediction value calculation unit 40 calculates the detection prediction value of the sensor A at the time point based on the prediction model 36 constructed for the sensor A and the detection value of the sensor A at a time point. In this way, the predicted value calculation unit 40 sequentially calculates the detected predicted value of each sensor.

The predicted value calculation unit 40 may construct the prediction model 36 by another method. For example, a prediction model 36 showing the correlation between the detection value of a specific sensor (for example, the sensor A) and the detection value of another sensor (for example, the sensor B) at the same time point is constructed based on the normal time detection value information. You may. According to the prediction model 36, it is possible to predict the detection prediction value of the specific sensor (sensor A) based on the detection values of the other sensors (sensor B).

For each sensor included in the sensor group 12, the abnormality degree calculating unit 42 compares the detection value of each sensor based on the difference between the detection predicted value predicted by the prediction value calculating unit 40 and the actual detection value of each sensor. The degree of abnormality (hereinafter, the degree of abnormality with respect to the detected value of the sensor will be simply referred to as “the degree of abnormality of the sensor”) is calculated. The abnormality degree calculation unit 42 calculates each sensor based on the predicted detection value of each sensor that is sequentially calculated and the actual detection value of each sensor that is sequentially detected by the sensor group 12 and that is sequentially transmitted from the central monitoring device 20. The abnormality degree of is calculated sequentially.

FIG. 3 shows the time change of the actual detection value and the detection prediction value of a certain sensor. In the graph of FIG. 3, the horizontal axis represents time, and the vertical axis represents the detection value or detection predicted value of the sensor. As described above, the detection value curve 50 is obtained based on the actual detection value sequentially transmitted from the central monitoring device 20, and the detection prediction value curve is obtained based on the detection prediction value sequentially calculated by the prediction value calculation unit 40. 52 is obtained. In addition, in FIG. 3, the time change of the detection value and the detection predicted value is indicated by a line, but the detection value and the detection predicted value are acquired intermittently, and the complementary processing is performed on the detection values at a plurality of time points. The detection value curve 50 is generated by performing the interpolation processing on the detection prediction values at a plurality of time points, and the detection prediction value curve 52 is generated. Further, the detection value curve 50 and the detection predicted value curve 52 do not necessarily have to be generated, and the set of the detection value and the detection predicted value at a plurality of time points over a certain period may be acquired.

The abnormality degree calculation unit 42 calculates the abnormality degree based on the difference between the predicted detection value and the detected value (indicated by d in FIG. 3 ). The degree of abnormality may be calculated based on the Euclidean distance between the predicted detection value and the detected value. Further, when the detection prediction value at a certain time point has a dispersed value, it may be calculated based on the Mahalanobis distance, which is the distance between the detection prediction value and the detection value in consideration of the dispersion.

By successively calculating the degree of abnormality by the degree-of-abnormality calculating unit 42, the degree-of-abnormality curve 54 showing the time-dependent change in the degree of abnormality is obtained as shown in FIG. In the graph of FIG. 4, the horizontal axis represents time and the vertical axis represents the degree of abnormality. In this way, the abnormality degree calculation unit 42 calculates the change over time in the abnormality degree of each sensor included in the sensor group 12. In addition, in FIG. 4, the temporal change of the degree of abnormality is shown by a line, but the degree of abnormality is intermittently acquired, and the abnormality degree curve 54 is obtained by complementing the abnormality degree at a plurality of time points. Is generated. Regarding the degree of abnormality, it is not always necessary to generate the abnormality degree curve 54, and the degree of abnormality may be calculated at a plurality of time points over a certain period of time.

The abnormality determination unit 44 determines whether or not the detected value of each sensor is an abnormal value based on the abnormality degree of each sensor included in the sensor group 12 calculated by the abnormality degree calculation unit 42. Specifically, an abnormality degree threshold value is set in advance for each sensor, and the abnormality determining unit 44 determines that the sensor outputs an abnormal value when the abnormality degree of a sensor is equal to or higher than the abnormality degree threshold value of the sensor. It is determined that it has been detected. The abnormality threshold value for each sensor may be stored in the storage unit 32.

The graph of FIG. 4 shows the abnormality threshold value s. In the example of FIG. 4, during the period when the abnormality degree indicated by the abnormality degree curve 54 is equal to or higher than the abnormality degree threshold value s, that is, from time t ₁ to time t ₂ , the abnormality determination unit 44 determines that the sensor is abnormal. It is determined that a value has been detected.

The abnormality importance degree calculation unit 46, when a certain sensor detects an abnormal value (hereinafter referred to as “at the time of abnormality detection”), the abnormality importance degree regarding the detection value of the sensor (hereinafter simply referred to as “sensor abnormality importance degree”). It is calculated). The abnormality importance is an index indicating whether or not the abnormal value detected by the sensor is due to a true malfunction in the device. Alternatively, the abnormality importance level is an index indicating which sensor, of the plurality of sensors, the service person should preferentially check when the plurality of sensors detect an abnormal value.

The abnormality importance degree calculating unit 46 determines the abnormality importance degree of the attention sensor based on not only the abnormality degree of the sensor of interest, which is the sensor for which the abnormality importance degree is calculated, but also the abnormality degree of another sensor other than the attention sensor. Calculate the degree. Here, the other sensor may be one sensor or a plurality of sensors. Hereinafter, the details of the process of the abnormality importance degree calculation unit 46 will be described, taking as an example the case where the sensor of interest is the sensor A and the other sensors are the sensor B and the sensor C.

FIG. 5A is an abnormality degree curve showing an example of the time variation of the abnormality degree of the sensor A, FIG. 5B is an abnormality degree curve showing an example of the time variation of the abnormality degree of the sensor B, and FIG. 5C is an abnormality of the sensor C. Anomaly curve showing an example of the degree of time change is shown. The vertical axis of each graph shown in FIGS. 5A to 5C represents the degree of abnormality and the horizontal axis represents time. In FIG. 5A, the abnormality degree indicated by the one-dot chain line indicates the abnormality degree threshold value s _A for the sensor A. Similarly, the Figure 5B, is shown in error probability threshold s _B is one-dot chain line relative to the sensor B, the same applies to FIG. 5C, the degree of abnormality threshold s _C is shown by the chain line relative to the sensor C. As shown in Figure 5A, the degree of abnormality of the sensor A is equal to or more than the abnormality threshold value of the degree of s _A from the time t _A is a time when the abnormality of the sensor A is equal to or greater than the abnormality threshold value of the degree of s _A, sensor This is a period up to time t′ _A when the abnormality degree of _A becomes less than the abnormality threshold s _A. That is, the period from the time t _A to the time t′ _A is the time when the abnormality of the sensor A is detected.

The abnormality importance level of the sensor A is based on the representative values of the abnormality levels of the sensor A, the sensor B, and the sensor C at the time of abnormality detection, that is, at a plurality of time points within the period from time t _A to time t′ _A. Calculated. In this embodiment, the average value is used as the representative value, but the representative value may be, for example, the median value or the maximum value. 5A shows the average value m _A (broken line) of the abnormality degree of the sensor A at the time of abnormality detection, and FIG. 5B shows the average value m _B (broken line) of the abnormality degree of the sensor B at the time of abnormality detection. Shows the average value m _C (broken line) of the abnormality degree of the sensor C at the time of abnormality detection.

More specifically, the abnormality importance calculator 46, the average value of the abnormality degree as to rather high abnormality importance of the sensor A large sensor A, the average sensor smaller the A of the abnormal temperature sensor A abnormal important Lower the degree. At the same time, the abnormality importance degree calculation unit 46 increases the abnormality importance degree of the sensor A as the average value of the abnormality degrees of the sensors B and C is smaller, and the sensor is larger as the average value of the abnormality degrees of the sensors B and C is larger. Lower the abnormal importance of A.

式１において、Ｉ_ＡはセンサＡの異常検出時におけるセンサＡの異常重要度である。また、ｋは他のセンサの個数を示す。本例では、他のセンサはセンサＢ及びセンサＣの２つであるから、ｋ＝２となる。 In the present embodiment, the abnormality importance level of the sensor A is the ratio of the average value of the abnormality level of the sensor A to the abnormality level threshold value s _A at the time of abnormality detection, and the average value of the abnormality level of the sensor B to the abnormality level threshold value s _B. It is calculated based on the ratio and the ratio of the average value of the abnormalities of the sensor C to the abnormality threshold s _C. Specifically, the abnormality importance level of the sensor A is calculated by the following formula.

In Formula 1, the I _A is abnormal importance of the sensor A at the time of abnormality detection of the sensors A. Further, k indicates the number of other sensors. In this example, since the other sensors are the sensor B and the sensor C, k=2.

The numerator on the right side of Expression 1 indicates the ratio of m _A (the average value of the abnormality degree of the sensor A at the time of abnormality detection) to s _A (the abnormality degree threshold value of the sensor A). Therefore, as the ratio of m _A for s _A m _A is smaller decreases, molecular decreases, i.e. abnormal importance I _A sensor A is reduced. Conversely, as the ratio of m _A is greater for s _A and m _A is increased, the molecular increases, i.e. abnormal importance I _A sensor A becomes large.

The first term of the denominator on the right side of Expression 1 indicates the ratio of m _B (the average value of the abnormality degree of the sensor B at the time of abnormality detection) to s _B (the abnormality degree threshold value of the sensor B). Similarly, the second term of the denominator on the right side of Expression 1 represents the ratio of m _C (average value of abnormality degree of sensor C at the time of abnormality detection) to s _C (threshold of abnormality degree of sensor C). Therefore, as m _B becomes smaller and the ratio of m _B to s _B becomes smaller, or as m _C becomes smaller and the ratio of m _C to s _C becomes smaller, the denominator becomes smaller, that is, the abnormality of the sensor A becomes smaller. The importance I _A becomes large. Conversely, as m _B increases and the ratio of m _B to s _B increases, or as m _C increases and the ratio of m _C to s _C increases, the denominator increases, that is, the sensor A The anomaly importance I _A becomes small.

In the present embodiment, the time when the abnormality of the sensor A is detected is set to the time t._AFrom time t'_AUp to the time t_AFrom time t'_AThe abnormality importance degree of the sensor A is calculated based on the average value of the abnormality degrees of the sensor A, the sensor B, and the sensor C in the period up to. That is, the abnormality importance degree calculation unit 46 determines that the time t'_AAlthough the abnormality importance degree of the sensor A is calculated later, the abnormality importance degree calculation unit 46 determines that the abnormality importance degree calculation unit 46 detects the abnormal value of the sensor A (time t in the example of FIG. 5A)._A), the degree of abnormality importance of the sensor A may be calculated. In this case, in equation 1, m_AAt time t _AAbnormality of sensor A at time point, m_BAt time t_AAbnormality of sensor B at time point, m_CAt time t_AThe abnormality degree I of the sensor A is replaced by the abnormality degree I of the sensor C at the time point._ACan be calculated. Thereby, the abnormality importance degree of the sensor A can be calculated earlier.

As shown in FIG. 5B, the average value m _B of the abnormality degree of the sensor B from the time t _A to the time t′ _A when the abnormality of the sensor A is detected is relatively high and is close to the abnormality threshold s _B. .. Further, as shown in FIG. 5C, the average value m _C of the abnormality degree of the sensor C from the time t _A to the time t′ _A is considerably high, and the value greatly exceeds the abnormality degree threshold value s _C. Therefore, in the example shown in FIGS. 5A to 5C, although the abnormality degree of the sensor A exceeds the abnormality degree threshold value s _A , the abnormality degree of the sensor B and the sensor C is high. The degree is relatively low.

FIG. 6A shows an abnormality degree curve showing the time variation of the abnormality degree of the sensor A, which is the same as FIG. 5A, FIG. 6B shows an abnormality degree curve showing another example of the time variation of the abnormality degree of the sensor B, and FIG. 6C. Shows an abnormality degree curve showing another example of the change over time of the abnormality degree of the sensor C, respectively. As can be seen by comparing FIGS. 5B and 6B, in the example of FIG. 6B, the average value m _B of the abnormality degree of the sensor B from the time t _A to the time t′ _A when the abnormality of the sensor A is detected is It is much smaller than the example of FIG. 5B. Similarly, as can be seen by comparing FIGS. 5C and 6C, in the example of FIG. 6C, the average value m _C of the abnormality degree of the sensor C from the time t _A to the time t′ _A is larger than that in the example of FIG. 5C. Is also quite small. Therefore, in the example shown in FIGS. 6A to 6C, the degree of abnormality importance of the sensor A is higher than that in the examples shown in FIGS. 5A to 5C.

Thus, the average value m _A degree of abnormality of the sensor A at the time of abnormality detection of the sensors A even if the same, depending on the degree of abnormality of the sensor B or sensor C, abnormal importance of the sensor A is changed obtain.

Thereby, the abnormality importance level of the sensor A can be an index indicating whether or not the abnormal value detected by the sensor A is due to a true malfunction in the device. For example, considering the case where all of the sensors A, B, and C are temperature sensors, when the sensor A detects an abnormal value, not only the temperature detected by the sensor A but also the temperatures detected by the sensors B and C rise. If so, the value of the abnormality importance of the sensor A is calculated to be relatively low. In such a case, the sensor A, that is, not a failure peculiar to the device installed on the 1st floor of the building 18, but a disturbance that commonly affects the sensor A, the sensor B, and the sensor C, such as a sudden rise in the outside temperature, causes the sensor A It is highly possible that all the detected temperatures of the sensor B, the sensor B, and the sensor C have risen. Therefore, it can be said that the lower the abnormality importance level of the sensor A, the lower the possibility that the abnormal value detected by the sensor A is due to the true malfunction of the device. On the other hand, when the sensor A detects an abnormal value, and the detected temperatures of the sensor B and the sensor C have not risen, a relatively high abnormality importance value of the sensor A is calculated. In such a case, the abnormal value detected by the sensor A is not due to the disturbance that affects the sensors A, B, and C in common, but is a problem specific to the device in which the sensor A is installed (that is, in the device). It is highly possible that a true defect) has occurred. Therefore, it can be said that the higher the abnormality importance of the sensor A is, the higher the possibility that the abnormal value detected by the sensor A is due to the true malfunction of the device.

Further, the abnormality importance level of the sensor A can be an index indicating which sensor, of the plurality of sensors, the service person should preferentially check when the plurality of sensors detect an abnormal value. .. For example, when the sensor A, the sensor B, and the sensor C simultaneously detect an abnormal value, the abnormality importance degree calculation unit 46 calculates the abnormality importance degree of each of the sensor A, the sensor B, and the sensor C. By presenting the abnormal importance degree of each sensor to the service person, the service person can preferentially perform confirmation work on the device provided with the sensor having the higher abnormal importance degree.

When the abnormality determination unit 44 determines that any of the sensors included in the sensor group 12 has detected an abnormal value, the abnormality notification unit 48 detects that an abnormal value has been detected, and the sensor that has detected the abnormal value is serviced. Notify the man. Further, the abnormality notifying unit 48 notifies the serviceman of the abnormality importance of the sensor that has detected the abnormal value, which is calculated by the abnormality importance calculating unit 46.

Specifically, the abnormality notifying unit 48 is triggered by the abnormality determining unit 44 determining that any of the sensors included in the sensor group 12 has detected the abnormal value, and detects the abnormal value. The detected sensor and abnormality information indicating the degree of abnormality importance of the sensor are push-transmitted to the serviceman terminal 16. As a result, the information included in the abnormality information is displayed on the display unit of the serviceman terminal 16. Alternatively, the abnormality notifying unit 48 may receive the request from the serviceman terminal 16 and transmit the abnormality information to the serviceman terminal 16.

FIG. 7 shows an example of an abnormal information display screen on the display unit of the serviceman terminal 16. FIG. 7 shows that three sensors, sensor A, sensor D, and sensor F, have detected an abnormal value. In the example of FIG. 7, since the sensor A, the sensor D, and the sensor F have the highest abnormality importance, the device on the first floor in which the sensor A is installed is prioritized to the service person. A message prompting you to check is displayed. That is, in the example of FIG. 7, when there are a plurality of sensors that have detected an abnormal value, the sensor with the highest abnormality importance is presented to the service person among the plurality of sensors. Alternatively, the abnormality importance of each sensor may be notified to the service person by a numerical value or the like.

The configuration outline of the abnormality importance degree calculation system 10 according to the present embodiment is as described above. The process flow of the server 14 will be described below with reference to the flowchart shown in FIG.

In step S10, the prediction value calculation unit 40 builds a prediction model for each sensor based on the normal-time detection value information stored in the past detection value information DB 34.

In step S<b>12, the server 14 receives the detection value information group including the detection values detected by the sensors included in the sensor group 12 from the central monitoring device 20.

In step S14, the prediction value calculation unit 40 calculates the detection prediction value of each sensor based on the prediction model of each sensor constructed in step S10 and the detection value of each sensor received in step S12.

In step S16, the abnormality degree calculation unit 42 calculates the abnormality degree of each sensor based on the difference between the detection value of each sensor received in step S12 and the detection predicted value of each sensor calculated in step S14. ..

In step S18, the abnormality determination unit 44 determines whether or not any of the sensors included in the sensor group 12 has detected an abnormal value. If all the sensors have not detected an abnormal value, the process returns to step S12 again, and the processes from step S12 to step S18 are repeated. That is, the prediction value calculation unit 40 sequentially calculates the detection prediction value of each sensor, and the abnormality degree calculation unit 42 sequentially calculates the abnormality degree of each sensor. If any sensor detects an abnormal value, the process proceeds to step S20.

In step S20, the abnormality importance calculating section 46, not only the abnormality of the sensor that detected the abnormal value, also based on the degree of abnormality of the other sensors other than the sensor, abnormality importance of the sensors detects an abnormal value To calculate.

In step S22, the abnormality determination unit 44, in addition to a sensor abnormality importance in step S2 0 is calculated, determines whether there is detected the sensor outliers. If there is, the process returns to step S20 again, and in step S20, the abnormality importance degree calculation unit 46 calculates the abnormality importance degree of the other sensor that has detected the abnormal value. When the degree of abnormality importance is calculated for all the sensors that have detected an abnormal value, the process proceeds to step S24.

In step S24, the abnormality notification unit 48 transmits to the serviceman terminal 16 the sensor that has detected the abnormal value and the abnormality information indicating the degree of importance of abnormality of the sensor. As a result, the abnormality information is displayed on the display unit of the serviceman terminal 16.

Although the embodiment according to the present invention has been described above, the present invention is not limited to the above embodiment, and various modifications can be made without departing from the spirit of the present invention.

10 abnormality importance calculation system, 12 sensor group, 14 server, 16 serviceman terminal, 18 building, 20 central monitoring device, 22 communication line, 30 communication unit, 32 storage unit, 34 past detection value information DB, 36 prediction model, 38 control unit, 40 predicted value calculation unit, 42 abnormality degree calculation unit, 44 abnormality determination unit, 46 abnormality importance degree calculation unit, 48 abnormality notification unit.

Claims (7)

A group of sensors provided in the device,
A server that receives a detection value detected by the sensor group,
An abnormality importance degree calculation system comprising:
The server is
For each sensor included in the sensor group, based on the difference between the predicted detection predicted value and the actual detection value, an abnormality degree calculation unit that calculates an abnormality degree for the detection value,
The abnormality degree with respect to the detection value of the attention sensor of the sensor group, the abnormality degree with respect to the detection value of the attention sensor at the time of abnormality detection, which is when the predetermined abnormality degree threshold value for the attention sensor or more, And an abnormality importance degree calculating unit that calculates an abnormality importance degree with respect to a detection value of the attention sensor, based on the abnormality degree with respect to a detection value of another sensor that is a sensor other than the attention sensor in the sensor group,
Have
Anomalous importance calculation system characterized by the following. The abnormality importance degree calculating unit increases the abnormality importance degree with respect to the detection value of the target sensor as the abnormality degree with respect to the detection value of the other sensor at the time of the abnormality detection is smaller, and the other abnormality degree at the time of the abnormality detection. The greater the degree of abnormality with respect to the detection value of the sensor, the lower the degree of abnormality importance with respect to the detection value of the sensor of interest,
The abnormality importance degree calculation system according to claim 1, wherein When the abnormality is detected, there is a period from the time when the abnormality degree with respect to the detection value of the attention sensor is equal to or higher than a predetermined abnormality degree threshold for the attention sensor to the time when the abnormality degree is less than the abnormality degree threshold. Then
The abnormality importance degree calculation unit, based on a representative value of the abnormality degree with respect to the detection value of the sensor of interest at a plurality of time points during the abnormality detection, and a representative value of the abnormality degree with respect to the detection value of the other sensor. Calculate the degree of abnormality importance for the detection value of the sensor of interest,
The abnormality importance degree calculation system according to claim 1, wherein Each sensor included in the sensor group is provided at a different point in the building,
The abnormality importance degree calculation system according to claim 1, wherein The server is
An abnormality notifying unit for notifying a person maintaining the device of abnormality information indicating that the detection value of the attention sensor has become an abnormal value, and abnormality information indicating an abnormality importance with respect to the detection value of the attention sensor
The abnormal importance degree calculation system according to claim 1, further comprising: For each sensor included in the sensor group provided in the device, based on the difference between the predicted detection predicted value and the actual detection value, an abnormality degree calculation unit that calculates an abnormality degree for the detection value,
The abnormality degree with respect to the detection value of the attention sensor of the sensor group, the abnormality degree with respect to the detection value of the attention sensor at the time of abnormality detection, which is when the predetermined abnormality degree threshold value for the attention sensor or more, And an abnormality importance degree calculating unit that calculates an abnormality importance degree with respect to a detection value of the attention sensor, based on the abnormality degree with respect to a detection value of another sensor that is a sensor other than the attention sensor in the sensor group,
An abnormality importance degree calculating device comprising: Computer,
For each sensor included in the sensor group provided in the device, based on the difference between the predicted detection predicted value and the actual detection value, an abnormality degree calculation unit that calculates an abnormality degree for the detection value,
The abnormality degree with respect to the detection value of the attention sensor of the sensor group, the abnormality degree with respect to the detection value of the attention sensor at the time of abnormality detection, which is when the predetermined abnormality degree threshold value for the attention sensor or more, And an abnormality importance degree calculating unit that calculates an abnormality importance degree with respect to a detection value of the attention sensor, based on the abnormality degree with respect to a detection value of another sensor that is a sensor other than the attention sensor in the sensor group,
Anomalous importance degree calculation program characterized by causing it to function as.

Images