JP7528046B2 - Anomaly detection device and anomaly detection method - Google Patents

Description

The present invention relates to an anomaly detection device and an anomaly detection method for detecting abnormal signs in equipment.

Early detection of abnormal operation signs in equipment installed in plants, office buildings, etc. is important for improving reliability, reducing downtime, and reducing running costs. The operating state of the equipment is managed by setting design values and thresholds based on experience for monitoring data such as temperature, pressure, and vibration acquired by sensors. For example, the condition monitoring system described in Patent Document 1 is a condition monitoring system that monitors the state of mechanical elements that constitute the equipment, and includes a sensor for detecting the state of the mechanical elements and a processing device for diagnosing abnormalities in the mechanical elements. The processing device calculates multiple feature amounts from the detection signal of the sensor, extracts at least one feature amount whose time-dependent trends are independent from each other as an effective feature amount from the calculated multiple feature amounts, and diagnoses abnormalities in the mechanical elements based on the extracted effective feature amount.

JP 2020-153836 A

The condition monitoring system described in Patent Document 1 sets a threshold based on statistics such as the average value and standard deviation of feature quantities, and judges an abnormality to exist when the feature quantity exceeds the threshold. However, because equipment behavior changes depending on the processing load and the environment (e.g., weather conditions), it may be difficult to detect early signs of abnormality in equipment operation using management based on thresholds based on past statistics, as in Patent Document 1. There is also a method for creating an estimated model of the operating state of each piece of equipment to judge an abnormality, but this has the problem of requiring a lot of work to create the estimated model.

The present invention was made in light of this background, and aims to provide an anomaly detection device and an anomaly detection method that enable efficient extraction of abnormal signs in equipment.

In order to solve the above-mentioned problems, the anomaly detection device of the present invention includes an abnormality sign index calculation unit that calculates an abnormality sign index that indicates the degree of abnormality of a state quantity of equipment during an analysis period based on statistics related to the state quantity of the equipment, which is time-series data during a learning period that is a period prior to an analysis period in which abnormalities in the equipment are detected, and a correlation analysis unit that calculates a correlation coefficient between the abnormality sign index and a candidate fluctuation factor that is a factor causing fluctuations in the abnormality sign index , wherein the equipment is a heat source system, and the candidate fluctuation factor is either an outside air temperature or a cooling load .

The present invention provides an anomaly detection device and an anomaly detection method that enable efficient extraction of abnormal symptoms in equipment. Problems, configurations, and effects other than those described above will become clear from the description of the embodiments below.

1 is an overall configuration diagram of a heat source system to be monitored by an anomaly detection device according to an embodiment of the present invention. 1 is a functional block diagram of an anomaly detection device according to an embodiment of the present invention. FIG. 4 is a data configuration diagram of a state quantity database according to the embodiment. FIG. 4 is a data configuration diagram of an exclusion period database according to the embodiment. FIG. 2 is a data configuration diagram of a state quantity attribute database according to the embodiment. FIG. 4 is a data configuration diagram of an abnormality sign index database according to the present embodiment. FIG. 2 is a data configuration diagram of an abnormality sign summary database according to the embodiment. 4 is a graph showing a cooling load of the refrigerator according to the embodiment. 4 is a graph showing a moving average of a chilled water supply temperature, a moving average of a chilled water return temperature, and a moving average of an outside air temperature of a chiller according to the present embodiment. 4 is a graph showing power consumption of the refrigerator according to the embodiment. 4 is a graph showing the power consumption of a chilled water pump, a cooling water pump, and a cooling tower according to the present embodiment. 4 is a graph showing abnormality symptom indicators of a system of the chiller according to the present embodiment. 5 is a graph showing abnormality sign indicators of a system of the chiller during operation according to the present embodiment. 10 is a graph showing an abnormality sign index of a system of a chiller when the chiller is not in operation according to the present embodiment. FIG. 2 is a screen configuration diagram of an analysis result display screen according to the embodiment. 4 is a flowchart of an abnormality sign analysis process according to the present embodiment. FIG. 11 is an overall configuration diagram of a heat source system according to a modified example of the present embodiment.

<Anomaly detection device: Overview>
An anomaly detection device in a form (embodiment) for carrying out the present invention will be described below. The anomaly detection device calculates and visualizes an anomaly sign index indicating deviation from an average value based on a feature amount calculated from a state amount of equipment measured by a sensor installed in the equipment. The feature amount is, for example, a moving average or moving standard deviation of the state amount. The anomaly detection device also calculates and visualizes a correlation coefficient between the anomaly sign index and a fluctuation factor. The fluctuation factors (candidate fluctuation factors) are, for example, time, outside air temperature, and cooling/heating load.

By looking at the abnormality sign indicators, the equipment manager (operator) can grasp signs of abnormality. In addition, by looking at the degree of correlation with variable factors, the manager can grasp the link with time changes, seasonal changes, and load changes, and obtain clues to clarify the cause of the abnormality. As a result, the manager will be able to grasp the possibility of an abnormality occurring, taking into account time, season, and load changes.

<Overall configuration of heat source system>
1 is an overall configuration diagram of a heat source system 200 to be monitored by an abnormality detection device 100 (see FIG. 2 described later) according to this embodiment. The heat source system 200 includes a cooling tower 211 and a chiller 221 of a first system, a cooling tower 212 and a chiller 222 of a second system, and a cooling tower 213 and a chiller 223 of a third system. The heat source system 200 sends cold water to a cold load 233 (such as an outdoor air conditioner or a fan coil unit) through a pipe and a chilled water supply header 232, and cools the cold water returned through a pipe and a chilled water return header 231. The three systems do not necessarily operate simultaneously, and when there are few outdoor air conditioners or fan coil units in operation and the load is small, only one system operates.

The heat source system 200 is equipped with sensors for measuring the temperature and flow rate of the chilled water, the power consumption of the cooling towers 211, 212, 213 and the chillers 221, 222, 223, etc., in order to monitor the operating status. In FIG. 1, a circled T indicates a temperature sensor, a circled F indicates a flow rate sensor, and a circled W indicates a power sensor. Note that circled △/▽ indicate pumps, which are also equipped with power sensors. These sensors transmit their measurement results to the anomaly detection device 100 at a predetermined timing, for example periodically.

<Anomaly detection device: Overall configuration>
2 is a functional block diagram of the anomaly detection device 100 according to the present embodiment. The anomaly detection device 100 is a computer, and includes a control unit 110, a storage unit 120, and an input/output unit 190. User interface devices such as a display, a keyboard, and a mouse are connected to the input/output unit 190. The input/output unit 190 also includes a communication device, and is capable of transmitting and receiving data to and from other devices including sensors installed in the heat source system 200. A media drive may also be connected to the input/output unit 190, enabling data exchange using a recording medium.

<Anomaly detection device: memory unit>
The storage unit 120 includes storage devices such as a read only memory (ROM), a random access memory (RAM), a solid state drive (SSD), etc. The storage unit 120 stores a state quantity database 130, an exclusion period database 140, a state quantity attribute database 150, an abnormal sign index database 160, an abnormal sign index upper limit value database 121, an abnormal sign summary database 170, and a program 128. The program 128 includes a description of an abnormal sign analysis process shown in FIG. 16, which will be described later.

Figure 3 is a data configuration diagram of the state quantity database 130 according to this embodiment. In the state quantity database 130, the measured values of the sensors are stored as state quantities (monitoring data) of the heat source system 200 together with the measurement date and time (acquisition date and time). The measured values include the power consumption of the main chillers 221, 222, and 223, the chilled water return temperature, the chilled water supply temperature, the cooling water return temperature, the cooling water supply temperature, the chilled water flow rate, the cooling water flow rate, the cooling load, and the COP (coefficient of performance, energy consumption efficiency). Other measured values include the power and frequency of the auxiliary chilled water pump, the cooling water pump, and the cooling towers 211, 212, and 213. The measured values are, for example, time-series data with a period of one hour. Although it is not a measured value of the heat source system 200 itself, the outside air temperature, which is a variable factor, is also stored in the state quantity database 130.

Figure 4 is a data configuration diagram of the exclusion period database 140 according to this embodiment. The exclusion period database 140 stores the reason for exclusion, start date and time, and end date and time for the period (exclusion period) that is excluded when calculating the abnormal sign index, along with identification information (referred to as ID in Figure 4). Reasons for exclusion include the start or stop sequence of equipment that shows transient behavior different from that during normal operation, and maintenance.

5 is a data configuration diagram of the state quantity attribute database 150 according to this embodiment. The state quantity attribute database 150 stores information related to the state quantities (monitoring data) of the heat source system 200, such as a name, an equipment classification, a system number, an equipment classification, an equipment type, a physical quantity type, and correlation calculations, in association with each other.
Names include "Refrigeration unit 1: power consumption" and "Refrigeration unit 1: chilled water supply temperature."
Equipment classifications include "heat source systems" and "air conditioning systems."

Equipment classifications include "main engine" and "auxiliary engine."
Equipment types include "refrigeration units,""chilled water pumps," and "cooling water pumps."
Physical quantity types include "electricity,""heat,""temperature," and "flow rate."
The correlation calculation is "Y" or "N" and specifies whether or not to calculate the correlation coefficient with the fluctuation factors of time, outside temperature, and cooling load.

6 is a data configuration diagram of the abnormal sign index database 160 according to this embodiment. The abnormal sign index database 160 calculates and stores abnormal sign indexes of the data stored in the state quantity database 130 (see FIG. 3 ). The data configuration of the abnormal sign index database 160 is similar to the data configuration of the state quantity database 130.
The abnormal sign index upper limit database 121 stores an upper limit value that is a criterion for determining whether an abnormal sign index is abnormal or not, in association with the item of the abnormal sign index.

FIG. 7 is a data configuration diagram of the abnormal sign summary database 170 according to this embodiment. The abnormal sign summary database 170 stores information related to abnormal sign indicators that exceed the upper limit value in the abnormal sign indicator upper limit value database 121, among the abnormal sign indicators in the abnormal sign indicator database 160. In more detail, the abnormal sign summary database 170 stores the name, equipment classification, system number, equipment classification, equipment type, physical quantity type, number of records of abnormal sign indicators, number of records exceeding the upper limit, maximum value of abnormal sign indicators, average value of abnormal sign indicators, and correlation coefficients with fluctuation factors (time, cooling load, and outside air temperature).

<Anomaly detection device: control unit>
2 , a description will be given of the configuration of the control unit 110 of the anomaly detection device 100. The control unit 110 is configured to include a CPU (Central Processing Unit) and is provided with a state quantity acquisition unit 111, a specified period exclusion unit 112, an operating state determination unit 113, a feature quantity calculation unit 114, an abnormal sign index calculation unit 115, a correlation analysis unit 116, and an analysis result display unit 117.
The state quantity acquisition unit 111 acquires measurement results (monitoring data, state quantities) of temperature, flow rate, and power consumption from sensors installed in the heat source system 200 (see FIG. 1 ) and stores them in the state quantity database 130. The state quantity acquisition unit 111 acquires the measurement results at a predetermined timing, for example, at one-hour intervals.

The specified period exclusion unit 112 excludes state quantities for a period in the excluded period database 140 when analyzing the state quantities in the state quantity database 130 .
The operation state determination unit 113 divides the state quantity of the period not excluded by the specified period exclusion unit 112 into an operation period and a non-operation period. In detail, the operation state determination unit 113 refers to the state quantity attribute database 150 (see FIG. 5 ) and determines the operation state based on the power consumption of the chillers 221, 222, and 223, which are main machines, for each system. For example, the operation state determination unit 113 determines that the operation state is in operation when the power consumption is equal to or greater than a predetermined value, and determines that the non-operation state is in operation when the power consumption is less than the predetermined value. The predetermined value is a value set for each chiller 221, 222, and 223. In addition, the operation state determination unit 113 determines the operation state of the auxiliary machines (chilled water pump, cooling water pump, cooling tower) corresponding to the main machines for each system according to the operation state of the main machines. Since equipment behaves differently depending on the operation state, abnormal signs can be detected with high accuracy by analyzing the state quantity according to the operation state.

The feature calculation unit 114 calculates feature quantities from the state quantities (monitoring data). Feature quantities include the state quantities themselves, which are time-series data, as well as the moving average and moving standard deviation of the state quantities over a predetermined period (e.g., one day). The moving standard deviation is the standard deviation of the moving average of the state quantities over a second predetermined period, and indicates the degree of variation in the moving average over the second predetermined period.

The abnormal sign index calculation unit 115 calculates the abnormal sign index and stores it in the abnormal sign index database 160 (see FIG. 6 ). In detail, the abnormal sign index calculation unit 115 calculates the average value μ and standard deviation σ of the feature values calculated by the feature value calculation unit 114 for each data item (item of state quantity), and calculates ((feature value-μ)/σ) ² as the abnormal sign index. Here, the average value μ is the average over the learning period and all systems (three systems in the heat source system 200). Note that the abnormal sign index calculation unit 115 may calculate the abnormal sign index for each data item and operation state.

The learning period is the period for finding the average value μ that serves as the reference for calculating the abnormality sign index, and is, for example, the period of state quantities stored in the state quantity database 130, or a period of one or two years. The analysis period is the period after the learning period, during which the abnormality sign index is calculated and the heat source system 200 is monitored. The default analysis period is after the learning period. The learning period and analysis period do not include the excluded period.

The abnormal sign index calculation unit 115 extracts abnormal sign index items (state quantity items, data items) that contain abnormal sign indexes that are equal to or greater than the upper limit in the abnormal sign index upper limit value database 121, and stores them in the abnormal sign summary database 170 (see FIG. 7). In detail, the abnormal sign index calculation unit 115 stores the name of the state quantity item from which the abnormal sign index was calculated, the system number, etc. in the "state quantity attribute" column of the abnormal sign summary database 170. The abnormal sign index calculation unit 115 stores the number of records (number of data) in the analysis target period, the number of records that exceed the upper limit, the maximum value of the abnormal sign index, and the average value in the "abnormal sign index" column. Below, the state quantities and abnormal sign indexes will be described with reference to FIG. 8 to FIG. 14.

Figure 8 is a graph 310 showing the cooling load of the chiller 221 according to this embodiment (see "Cold Load" of "Chiller 1" in Figure 3). Graph 310 shows data for three days, during which the chiller 221 is stopped at night (from 9 pm to 7 am the following day), and the cooling load is almost zero. Note that graphs 310, 320, 330, and 340 of the moving averages of the cooling load, temperature, and power shown in Figures 8 to 11 are moving average graphs, and will be described without the term "moving average". For example, "outside air temperature" refers to the moving average of the outside air temperature.

Figure 9 is a graph 320 showing the moving average of the chilled water supply temperature (plotted with ●), the moving average of the chilled water return temperature (plotted with ◯), and the moving average of the outside air temperature (plotted with ▲) of the chiller 221 according to this embodiment. The outside air temperature is not a state quantity of the equipment, but is listed for reference. It can be seen that the chilled water supply temperature and the chilled water return temperature approach the outside air temperature at night (they move according to the flow). The chilled water supply temperature during the day is almost constant and as set. On the other hand, the chilled water return temperature is high in the early afternoon (from 12:00 to 15:00), and is thought to be influenced by the outside air temperature.

Figure 10 is a graph 330 showing the power consumption of the refrigerator 221 according to this embodiment. As with graph 310 (see Figure 8), the refrigerator 221 is stopped at night (from 9 pm to 7 am the following day), and the power consumption is almost zero.

Figure 11 is a graph 340 showing the power consumption of the chilled water pump (plotted with ●), the cooling water pump (plotted with ◯), and the cooling tower 211 (plotted with ▲) according to this embodiment. With some exceptions (see the power consumption of the chilled water pump from 9 pm to midnight on the third day (plotted with ●)), power consumption is lower at night, and the chilled water pump, cooling water pump, and cooling tower are not operating.

Figure 12 is a graph 350 showing an abnormality sign index of the chiller 221 system according to this embodiment. The abnormality sign index of the chiller 221 system is the maximum value of the abnormality sign index having the moving average of each of the chiller 221 cooling water supply temperature, the chiller 221 cooling water return temperature, the power consumption of the chiller 221, the power consumption of the chiller 221 chilled water pump, the power consumption of the chiller 221 cooling water pump, and the power consumption of the chiller 221 cooling tower as a feature quantity. If there are other state quantities related to the chiller 221 system, the maximum value of the abnormality sign index may include the abnormality sign index having the moving average of the other state quantities as a feature quantity.

The abnormality symptom index in the area 351 enclosed by the dotted line shows a high value, but this is a value immediately after inspection and maintenance and is not abnormal. In order to remove such high abnormality symptom indexes that are not abnormal, the specified period exclusion unit 112 excludes the state quantities for the period stored in the excluded period database 140 (see FIG. 4) from the analysis target.

Figure 13 is a graph 360 showing abnormality symptom indicators for the chiller 221 system during operation according to this embodiment. The abnormality symptom indicators in the area 361 enclosed by the dotted line indicate high values. During the same time period, the chilled water supply temperature of the chiller 221 (● in graph 320 (see Figure 9)) has risen, which may indicate that the supply air temperature of the air conditioning system has risen.

14 is a graph 370 showing abnormality symptom indicators of the chiller 221 system when not in operation according to this embodiment. The abnormality symptom indicators in the area 371 surrounded by a dotted line indicate high values. The power consumption of the chilled water pump during the same time period (● in graph 340 (see FIG. 11)) is also increasing. Although it is normally not in operation at night, the chilled water pump is operating, which may be causing a loss of power.
By calculating the abnormality sign index in this manner, it becomes possible to grasp the possibility of an abnormality in the heat source system 200 that would be difficult to grasp using only the feature quantities (graphs 310, 320, 330, and 340 shown in FIGS. 8 to 11).

Returning to Fig. 2, the explanation of the control unit 110 will be continued. The correlation analysis unit 116 calculates a correlation coefficient between an abnormal sign index and a fluctuation factor (time, outside air temperature, cooling/heating load, etc.) for a state quantity (data item) for which the "correlation calculation" in the state quantity attribute database 150 (see Fig. 5) is "Y".
By calculating the correlation coefficient with time (for example, Pearson's product-moment correlation coefficient), the change trend of the status quantity item (tendency to increase/no change/tendency to decrease) can be obtained. By calculating the correlation coefficient with outside temperature, the seasonality of the status quantity item and its relationship with air temperature can be obtained. In addition, by calculating the correlation coefficient with cooling and heating load, the load linkage of the status quantity item can be obtained. By looking at the degree of correlation with fluctuation factors, managers can understand the linkage with time changes, seasonal changes, and load changes, and can obtain clues to clarify the cause of the abnormality.

The analysis result display unit 117 displays an analysis result display screen 400 (see FIG. 15 described later) on a display connected to the input/output unit 190. FIG. 15 is a screen configuration diagram of the analysis result display screen 400 according to this embodiment.
The maximum abnormality sign index display area 410 displays the maximum value of the abnormality sign index in the analysis result display period included in the analysis target period as a graph. The default analysis result display period is, for example, the most recent three months. The hatched period indicates the period during which the maximum value of the abnormality sign index exceeded a predetermined threshold.

The analysis result display period is displayed in the analysis result display area 420. The analysis result display period is editable. When the "Reflect" button is pressed after editing the analysis result display period, a graph of the edited period is displayed in the maximum abnormal sign indicator display area 410 and the abnormal sign indicator display area 450 described below.

The period length change specification area 430 contains buttons for changing the length of the analysis result display period. For example, pressing the "-1 day" button shortens the analysis result display period by one day. The start date change specification area 440 contains buttons for changing the start date of the analysis result display period. For example, pressing the "Previous day" button shortens the start date of the analysis result display period by one day.

A graph of the abnormal sign index and the original data (the state quantity itself or the moving average of the state quantity) of each data item during the analysis result display period is displayed in the abnormal sign index display area 450. In Fig. 15, the abnormal sign index and the original data of "chiller 1: power consumption" and "chiller 1: cooling load" are displayed.
The hatched period in the graph for "Refrigerator 1: Power consumption" is the same as the hatched period in the graph for the maximum value of the abnormality sign index. This shows that the abnormality sign index for "Refrigerator 1: Power consumption" exceeds a predetermined threshold value related to the maximum value of the abnormality sign index.

When the "Abnormal Sign Summary" button 481 is pressed, the abnormal sign summary database 170 (see FIG. 7) is displayed. When the "Exclusion Period Setting" button 482 is pressed, the contents of the exclusion period database 140 (see FIG. 4) are displayed. The exclusion period database 140 is updated by editing the contents.

<Abnormality Sign Analysis Processing>
FIG. 16 is a flowchart of the abnormality sign analysis process according to this embodiment.
In step S11, the feature amount calculation unit 114 calculates the feature amount of each state amount during a preset learning period, and the abnormal sign index calculation unit 115 calculates the average value μ and standard deviation σ of the feature amounts.
In step S12, the abnormal sign index calculation unit 115 sets the analysis target period to a default period. The default period is a period after the learning period.

In step S13, the feature amount calculation unit 114 calculates the feature amount for the analysis target period.
In step S14, the abnormal sign index calculation unit 115 calculates an abnormal sign index from the feature amount calculated in step S13, and stores the result in the abnormal sign index database 160. The abnormal sign index calculation unit 115 also stores in the abnormal sign summary database 170 state quantity attributes, the number of records, and the like related to the items of the abnormal sign index that exceed the upper limit value stored in the abnormal sign index upper limit value database 121.

In step S15, the correlation analysis unit 116 calculates a correlation coefficient between the abnormal sign index and the fluctuation factor in the analysis target period. The correlation analysis unit 116 stores the correlation coefficient in the abnormal sign summary database 170.
In step S16, the analysis result display unit 117 displays the analysis result display screen 400 (see FIG. 15).
In step S17, the analysis result display unit 117 branches and proceeds according to the button pressed in the analysis result display screen 400. When a button related to changing the analysis result display period arranged in the period length change designation area 430 and the start date change designation area 440 is pressed (step S17 → change), the analysis result display unit 117 proceeds to step S18.

When the "Reflect" button arranged in the analysis result display area 420 is pressed (step S17->reflect), the analysis result display unit 117 proceeds to step S19.
When the "set exclusion period" button 482 is pressed (step S17->set exclusion period), the analysis result display unit 117 proceeds to step S20.
When the "Abnormal Sign Summary" button 481 is pressed (step S17->Abnormal Sign Summary), the analysis result display unit 117 proceeds to step S21.
When the "End" button 483 is pressed (step S17->End), the analysis result display unit 117 ends the abnormal sign analysis process.

In step S18, the analysis result display unit 117 displays a graph of the analysis result display period corresponding to the button pressed in step S17, and returns to step S17. In detail, the analysis result display unit 117 displays a graph of the analysis result display period showing the maximum value of the abnormal sign index in the maximum abnormal sign index display area 410, and displays a graph of the abnormal sign index of each data item and the analysis result display period of the original data in the abnormal sign index display area 450. For example, when the "+1 day" button located in the period length change specification area 430 is pressed, the analysis result display unit 117 extends the analysis result display period by one day and displays it.

In step S19, the analysis result display unit 117 displays a graph of the maximum abnormal sign index, the abnormal sign index of each data item, and the analysis result display period after editing of the original data, and then returns to step S17.
In step S20, the analysis result display unit 117 displays the contents of the exclusion period database 140, accepts any changes to the exclusion period, and reflects these changes in the exclusion period database 140. Then, the analysis result display unit 117 returns to step S11.
In step S21, the analysis result display unit 117 displays the contents of the abnormal sign summary database 170, and the process returns to step S17.

16, after the abnormality sign index is calculated, the analysis result display unit 117 displays a graph for a specified display period. When a new state quantity is acquired, an abnormality sign index related to the state quantity may be calculated and reflected in the graph.
16, the average and standard deviation of the feature quantities and the abnormal sign index are calculated each time the abnormal sign analysis process is performed. The analysis result display unit 117 may store these calculation results and display the graphs using the stored calculation results unless the learning period or the exclusion period is changed.

<Features of the anomaly detection device>
The anomaly detection device 100 calculates an abnormality sign index indicating deviation from an average value based on the feature quantities of the state quantities of the heat source system 200 measured by sensors installed in the heat source system 200, and displays the analysis result display screen 400. The anomaly detection device also calculates and visualizes a correlation coefficient between the abnormality sign index and a fluctuation factor. Examples of the fluctuation factors include time, outside air temperature, and cooling/heating load.

By looking at the abnormality sign indicators, the equipment manager (operator) can grasp signs of abnormality. In addition, by looking at the degree of correlation with variable factors (correlation coefficient), the manager can grasp the link with time changes, seasonal changes, and load changes. As a result, the manager will be able to grasp the possibility of an abnormality occurring, taking into account time, season, and load changes.

<<Variation: Abnormality Sign Indicator>>
In the above embodiment, the abnormal sign index is calculated for each item of the state quantity (see abnormal sign index database 160 in FIG. 6). The maximum value of the abnormal sign indexes of the data items related to an individual piece of equipment (devices constituting the equipment), for example, the refrigerator 221, may be used as the abnormal sign index for the refrigerator 221 (see graphs 350, 360, and 370 in FIGS. 12 to 14).

In the above embodiment, the abnormal sign index is calculated based on the average value μ and standard deviation σ of the feature of the state quantity during the learning period. The abnormal sign index may be calculated using other methods. For example, the feature of the state quantity may be classified into one or more clusters by cluster analysis, and the abnormal sign index may be calculated based on the clusters. For example, the abnormal sign index calculation unit 115 may generate clusters from data during the learning period, and the distance between the data of the analysis target period and the closest cluster may be used as the abnormal sign index. Furthermore, the abnormal sign index calculation unit 115 may perform clustering using the k-means method, which standardizes multiple data items and calculates the Euclidean distance, and perform an abnormality determination using the k-nearest neighbor method.

<<Variation: Operation Mode>>
An operation mode may be set in the calculation of the average value μ and the standard deviation σ and the calculation of the abnormal sign index during the learning period. The operation mode is a combination of operation/non-operation of the refrigerators 221, 222, and 223, such as an operation mode in which the refrigerator 221 is operating and the refrigerators 222 and 223 are not operating, and an operation mode in which all the refrigerators 221, 222, and 223 are operating. The abnormal sign index calculation unit 115 calculates the average value μ and the standard deviation σ for each operation mode. Furthermore, when calculating the abnormal sign index, the abnormal sign index calculation unit 115 uses the average value μ and the standard deviation σ according to the operation mode.
In the heat source system 200 of the embodiment described above, the chillers 221, 222, and 223 are connected one-to-one with the cooling towers 211, 212, and 213 through cooling water piping. Alternatively, there is a configuration in which the front chillers are connected to all the cooling towers through cooling water piping.

Figure 17 is an overall configuration diagram of a heat source system 200A according to a modified example of this embodiment. The chillers 221, 222, 223 and the cooling towers 211, 212, 213 are connected by cooling water piping via a cooling water return header 241 and a cooling water supply header 242. In the case of such a heat source system 200A, the cooling water piping (cooling water system) is shared by multiple chillers and cooling towers, so the operating status of the entire heat source equipment cannot be distinguished only by the operating status of each chiller. For this reason, accuracy is improved by taking into account the operating status (operating mode) of all the chillers 221, 222, 223 and calculating the abnormality sign index according to each operating status.

Other variations
Although several embodiments of the present invention have been described above, these embodiments are merely illustrative and do not limit the technical scope of the present invention. For example, in the above-described embodiments, the anomaly detection device 100 stores various databases in the storage unit 120, but the anomaly detection device 100 may access a database in another device, such as a cloud storage device.
The present invention can take various other embodiments, and furthermore, various modifications such as omissions and substitutions can be made without departing from the gist of the present invention. These embodiments and modifications are included in the scope and gist of the invention described in this specification, etc., and are included in the scope of the invention described in the claims and their equivalents.

REFERENCE SIGNS LIST 100 Abnormality detection device 111 State quantity acquisition unit 112 Designated period exclusion unit 113 Operation state determination unit 115 Abnormal sign index calculation unit 116 Correlation analysis unit 117 Analysis result display unit 130 State quantity database 140 Exclusion period database 150 State quantity attribute database 160 Abnormal sign index database 121 Abnormal sign index upper limit value database 170 Abnormal sign summary database 200, 200A Heat source system

Claims (8)

an abnormality sign index calculation unit that calculates an abnormality sign index that indicates the degree of abnormality in a state quantity of the equipment during an analysis period based on a statistical quantity related to the state quantity of the equipment, which is time-series data during a learning period that is a period prior to the analysis period in which an abnormality in the equipment is detected;
a correlation analysis unit that calculates a correlation coefficient between the abnormal sign index and a candidate fluctuation factor that is a fluctuation factor of the abnormal sign index ,
The facility is a heat source system,
The candidate fluctuation factors are either the outside air temperature or the cooling load.
An anomaly detection device comprising: The abnormality sign index calculation unit
calculating the abnormality sign index based on a ratio of a deviation of the feature quantity of the state quantity from an average value and a standard deviation of the feature quantity;
The feature amount is
The anomaly detection device according to claim 1 , wherein the state quantity is any one of the state quantity itself, a moving average of the state quantity, and a moving standard deviation of the state quantity. The abnormality sign index calculation unit is
classifying the feature quantities of the state quantities into clusters and calculating the abnormal sign indicators based on the clusters;
The feature amount is
The anomaly detection device according to claim 1 , wherein the state quantity is any one of the state quantity itself, a moving average of the state quantity, and a moving standard deviation of the state quantity. The abnormality sign index calculation unit is
2. The anomaly detection device according to claim 1, wherein a maximum value of the anomaly sign indexes of state quantities related to the devices constituting the facility is calculated as the anomaly sign index of the devices. The facility is a heat source system having a plurality of systems,
The abnormality detection device according to claim 1 , wherein the abnormality sign index calculation unit calculates the abnormality sign index for each combination of operating states in each of the plurality of systems. The anomaly detection device according to claim 1, characterized in that the state quantities include at least one of the following: power, chilled water return temperature, chilled water supply temperature, cooling water return temperature, cooling water supply temperature, chilled water flow rate, cooling water flow rate, cooling load, and energy consumption efficiency of a chilled water pump which is an auxiliary unit of the heat source system; power and frequency of a cooling water pump which is an auxiliary unit; and power and frequency of a cooling tower which is an auxiliary unit. The anomaly detection device according to claim 1 , further comprising an analysis result display unit that displays a maximum abnormality sign index at each time among the abnormality sign indexes. An anomaly detection method for an anomaly detection device, comprising:
The abnormality detection device includes:
A step of calculating an abnormality sign index that indicates the degree of abnormality in the state quantity of the equipment during an analysis period based on a statistical quantity related to the state quantity of the equipment, which is time-series data during a learning period that is a period prior to the analysis period in which an abnormality in the equipment is detected;
Executing a step of calculating a correlation coefficient between the abnormal sign index and a candidate fluctuation factor that is a fluctuation factor of the abnormal sign index ;
The facility is a heat source system,
The candidate fluctuation factors are either the outside air temperature or the cooling load.
Anomaly detection methods.

Images