JP6438124B2 - Operation management system and operation management method - Google Patents

Description

  The present invention relates to an operation management system and an operation management method, and is suitably applied to an operation management system and an operation management method for managing the operation of an apparatus installed in a large-scale wide-area multi-site.

  In recent years, sensor data such as operation data is gathered at a central monitoring base from devices installed at many bases distributed on a large scale, and big data is utilized. For example, in the central monitoring base, a sign such as an abnormal state or a failure state of an apparatus installed in a large number of bases is detected in advance based on the collected data.

  For example, in Patent Document 1, in a facility management apparatus, a set of element models for each component of the facility is used as a facility model, and the facility model of each facility is periodically sucked up by the central monitoring device, and a similar element model from the facility model. Is extracted and stored as a similar element model group. When a component is designated by the occurrence of a certain event, the central monitoring apparatus detects a similar element model group including the designated element model and detects a sign such as a failure. .

  Moreover, in patent document 2, in the air-conditioning heat load prediction system, a prediction model that takes into account the comparison results with the temporary prediction model based on past environmental element data and simulated data and the special characteristics of the target building is generated. Predicting an accurate air conditioning heat load prediction value is performed.

International Publication No. 01/069338 JP-A-8-35796

  However, in Patent Document 1 and Patent Document 2, since the sign detection process and the prediction process described above are performed centrally at the central base, traffic increases excessively due to the transmission / reception of a large number of operation data as the number of bases increases. There has been a problem that scalability cannot be guaranteed due to an increase in the number of bases.

  The present invention has been made in consideration of the above points. When detecting abnormal signs by aggregating operation data of a large number of target sites at a central site, data transfer is performed by performing a part of detection processing at each site. Reduce volume and increase computational resources at the central location.

  In order to solve such a problem, in the present invention, an operation management system for supporting operation management by monitoring the operating status of devices installed at a plurality of target sites in a central site, the plurality of target sites, An operation sensor for collecting operation data for monitoring the operation status of the apparatus, a storage unit for storing operation data collected by the operation sensor, a collection item of the operation data, and a collection frequency are determined, and the operation is performed. A control unit that transfers data to the central site, wherein the central site stores a storage unit that stores the operational data, and operating states of devices installed at the plurality of target sites based on the operational data A control unit that determines the abnormality, and the control unit at the target site detects an abnormality of the device from the operation data based on a predetermined condition defined in the design specification of the device. Based on the predictive alert preliminary alert rule for preliminarily detecting the abnormal sign specified at the central location, the abnormal sign is detected from the operation data and the preliminary sign detection preliminary alert is generated. The control unit of the central base constructs a model of the device for sign detection for detecting a sign of abnormality based on operation data of the device collected from the plurality of target bases. Based on the model of the device, generate the predictive detection preliminary alert rule, execute the predictive analysis to analyze the predictive sign of abnormality utilizing the model of the constructed device and the operation data of the target device of the target base, An operation management system is provided, wherein an abnormality cause analysis is performed for analyzing a cause of an abnormality based on the operation data of the target device at the target base.

  Further, in order to solve such a problem, the present invention provides an operation management method in an operation management system that monitors the operation status of devices installed at a plurality of target sites in a central site and supports operation management. The base control unit determines the operation data collection item and the collection frequency for monitoring the operation status of the device, and transfers the operation data to the central base, and the central base control unit, Constructing a model of the device for sign detection for detecting a sign of abnormality based on operation data of the device collected from the plurality of target bases, and the control unit of the central base comprising the construction Generating a predictive detection preliminary alert rule for preliminarily detecting an abnormal sign based on the model of the device, and the control unit of the target base A warning detection preliminary alert is generated from the operation data based on the warning detection preliminary alert rule specified by a point, and a control unit at the central site detects the warning detection preliminary alert. When it occurs, performing a predictive analysis that analyzes a predictor of abnormality utilizing the operation data of the target device of the target device and the target device of the target base, the control unit of the target base, A step of detecting an abnormality of the apparatus from the operation data and generating an abnormality detection alert based on a predetermined condition defined in the design specification of the apparatus, and the control unit of the central base has generated the abnormality detection alert Or the cause of the abnormality based on the operation data of the target device at the target site when it is determined that there is an abnormal sign as a result of the detailed analysis of the sign Characterized in that it comprises the steps of performing an abnormality cause analysis to analyze the operation management method is provided.

  According to the present invention, when the operation data of a large number of target bases are aggregated at the central base and an abnormality sign is detected, a part of the detection processing is performed at each base, so Scalability required by increasing data capacity can be ensured.

It is a conceptual diagram which shows the data analysis process implemented with the operation management system which concerns on one Embodiment of this invention, and the state transition of each process. It is a block diagram which shows the structure of the operation management system concerning the embodiment. It is a block diagram which shows the structure of the operation management system concerning the embodiment. It is a chart which shows an example of the sensor data for learning concerning the embodiment. It is a graph which shows an example of the sensor data for predictive analysis concerning the embodiment. It is a graph which shows an example of the sensor data for abnormality analysis concerning the embodiment. It is explanatory drawing explaining the operation management process in the normal state concerning the embodiment. It is explanatory drawing explaining the operation management process in the abnormality detection state concerning the embodiment. It is explanatory drawing explaining operation management processing when the sign detection preliminary alert concerning the embodiment occurs. It is explanatory drawing explaining operation management processing when the sign detection preliminary alert concerning the embodiment occurs. It is explanatory drawing explaining the operation management process when it is judged that there is no abnormality sign concerning the embodiment. It is explanatory drawing explaining operation management processing when it is judged that there is an abnormal sign concerning the embodiment.

  Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

(1) Outline of the present embodiment First, the outline of the present embodiment will be described with reference to FIG. FIG. 1 is a conceptual diagram showing data analysis processing performed in the operation management system according to the present embodiment and state transition of each processing. In the present embodiment, the operation data of a large number of target bases dispersed on a large scale are collected at the central base, and abnormal (failure) sign detection processing is performed using the operation data (big data).

  As shown in FIG. 1, the operational data sent from each base to the central base includes “learning sensor data” for building a model for predictive detection, and detailed prediction analysis for determining whether there is an abnormal sign. For example, “predictive analysis sensor data” and “abnormality analysis sensor data” for analyzing the cause of the abnormality.

  The learning sensor data used for the model construction predictor analysis 10 is data transmitted from each base to the central base at a normal time, and has a lower recording frequency (for example, every 30 minutes) than the predictor analysis sensor data described later. 1 time) Sensor data.

  In the model construction predictor analysis 10, a model including a feature amount, a threshold value, and the like for preliminarily determining a sign of abnormality of the apparatus is constructed at the central base based on the learning sensor data collected at the central base (STEP 10 ). In the model construction predictor analysis 10, whether or not there is a possibility of abnormality sign detection is analyzed using the learning sensor data at the central base, and if there is a possibility, it is set as a sign detection preliminary determination (STEP 11). Further, in the model construction predictive analysis 10, based on the predictive detection preliminary alert rule set based on the model of each device for predictive detection, the predictive detection preliminary alert is raised at each base (STEP 12). Also, an abnormality detection alert is raised at each site based on the conditions defined in the design specifications of each device (STEP 13).

  The sensor data for predictive analysis for predictive detailed analysis 20 for determining whether there is an abnormal sign has the possibility of detecting an abnormal sign when a preliminary detection warning is generated in STEP 12 and by preliminary detection in STEP 11 Is the same data item as the learning sensor data described above, but the recording frequency is higher than the learning sensor data (for example, once every 30 seconds). ) Sensor data.

  In the predictor detailed analysis 20, sensor data for predictor analysis is accumulated at the central base (STEP 20), and it is analyzed whether there is an anomaly sign or not (STEP 21) and whether there is an anomaly sign (STEP 22). . Further, when an abnormality is detected at each site, an abnormality detection alert is issued (STEP 23). The process for raising the abnormality detection alert in STEP 23 is the same as the process in STEP 13 described above.

  The sensor data for abnormality analysis for analyzing the cause of abnormality 30 is transmitted from each base to the central base when an abnormality detection alert occurs in STEP 23 or when an abnormality detection alert occurs in STEP 13. This is data obtained by adding a valid data item to identify the cause of abnormality to the above-described sensor data for predictive analysis.

  In the abnormality cause analysis 30, sensor data for abnormality analysis is accumulated (STEP 31), and the cause of the abnormality of the apparatus is specified (STEP 32).

  In the present embodiment, a part of the analysis process described above is executed at each site, and each data necessary for the analysis is transmitted at a suitable timing. Specifically, the processing of STEP 12 described above is performed at each site. That is, in each base, the operation data of each device in the base is monitored, and the sign detection preliminary alert is generated based on the sign detection preliminary alert rule for generating the sign detection preliminary alert. The sign detection preliminary alert rule is set at the central base and transmitted from the central base to each base.

  The central base usually collects sensor data for learning with a low recording frequency from each base, generates a preliminary warning detection alert rule including a characteristic amount and threshold for predictive detection preliminary judgment, For improving the feature amount and threshold. When the central site determines that there is a possibility of predictive detection based on the preliminary detection, or when a preliminary detection alert is generated at any of the sites, Collect sensor data for predictive analysis with higher recording frequency than sensor data for learning from the base. In addition, when an abnormality detection alert occurs based on the conditions defined in the design specifications of each device at each site, the central site has a data item that is effective for identifying the cause of the abnormality in the sensor data for predictive analysis. Collect the added data for abnormality analysis.

  As described above, the capacity of the three types of data, that is, the sensor data for learning, the sensor data for predictive analysis, and the sensor data for abnormal analysis, has the largest capacity of the data for abnormal analysis, and then the sensor data for predictive analysis and the sensor data for learning It becomes the order of. Therefore, the data sent from each site to the central site is usually sent with the smallest sensor data for learning, and the sensor data for predictive analysis is sent when there is a preliminary detection detection or preliminary detection alert. The sensor data for abnormality analysis is transmitted when it is determined that there is a sign of abnormality or when an abnormality detection alert is raised.

  In this way, a part of the detection process is performed at each base, and the sign analysis sensor data and abnormality analysis sensor data having a larger capacity than the learning sensor data are transmitted based on the sign detection result performed at each base. As a result, it is possible to construct an operation management system for monitoring anomalies at a large number of bases that secures the scalability required due to an increase in the number of target bases and an increase in data capacity by reducing communication traffic on the network and the like.

(2) Configuration of Operation Management System Next, the configuration of the operation management system will be described with reference to FIGS. 2 and 3. As shown in FIG. 2, in the operation management system, a central site 200 and target sites 100A, 100B, 100C, and 100D are connected via a network 300. In FIG. 2, four target bases 100A, 100B, 100C, and 100D are shown, but in reality, the central base 200 and a large number of target bases are connected. In the following description, the target bases 100A, 100B, 100C, and 100D are collectively referred to as the target base 100.

(2-1) Hardware Configuration of Each Device As shown in FIG. 2, the target site 100 includes a control unit 101, a temperature sensor (T in the figure) 102, a humidity sensor (H in the figure) 103, and an atmospheric pressure sensor (in the figure). P) 104, BMS (Building Management System) 105, operation sensor 106, and a plurality of devices (device A and device B in the figure) 107.

  The control unit 101 of the target site 100 includes information processing resources such as a CPU and a memory. The CPU functions as an arithmetic processing device and controls the operation of the target base 100 according to a program, arithmetic parameters, etc. stored in the memory.

  A temperature sensor (T in the figure) 102 is a sensor for measuring temperature. A humidity sensor (H in the figure) 103 is a sensor that measures humidity. The atmospheric pressure sensor (P in the figure) 104 is a sensor that measures atmospheric pressure.

  A BMS (Building Management System) 105 performs automation of each device 107 of the base 100 and monitoring control of various functions. The BMS 105 monitors each device and notifies an abnormality detection alert to the data transfer control based on the conditions defined by the design specifications of each device. The operation sensor 106 is a sensor that detects an operation state of the device 107. A plurality of devices (device A and device B) 107 are heating ventilating air conditioning (HVAC) such as heating, ventilation, and air conditioning.

  In addition, the control unit 101 includes a communication interface including a communication device for connecting to a network. Examples of the communication device include those that support IP-VPN, wide area Ethernet, dedicated lines, mobile communication such as 3G and 4G. This communication apparatus transmits / receives various data to / from the central base 200 via a network.

  The base 100 includes an information input device such as a keyboard, a switch, a pointing device, and a microphone, and an information output device such as a monitor display and a speaker.

  The central site 200 includes information processing resources such as a CPU and a memory. The CPU functions as an arithmetic processing device, and controls the operation of the central base 200 according to programs, arithmetic parameters, and the like stored in the memory.

  In addition, the central site 200 includes a communication interface including a communication device for connecting to a network. Examples of the communication device include those that support IP-VPN, wide area Ethernet, dedicated lines, mobile communication such as 3G and 4G. This communication apparatus transmits / receives various data to / from the base 100 via a network.

  The central base 200 includes an information input device such as a keyboard, a switch, a pointing device, and a microphone, and an information output device such as a monitor display and a speaker.

  Further, the central site 200 may include a storage device for storing data. The storage device includes a storage medium, a recording device that records data on the storage medium, a reading device that reads data from the storage medium, and a deletion device that deletes data recorded on the storage medium. The storage device is composed of, for example, an HDD (Hard Disk Drive), drives a hard disk, and stores programs executed by the CPU and various data.

(2-2) Software Configuration of Each Device Next, the software configuration of each device will be described. As shown in FIG. 2, examples of programs executed by the control unit 101 include policy base control 110, data transfer control 120, alert rule monitoring 130, and sensor data collection control 140.

  The policy base control 110 issues various commands to the data transfer control 120, the alert rule monitoring 130, and the sensor data collection control 140 based on various policies. The data transfer control 120 controls the transfer of various data. The alert rule monitor 130 monitors sensor data collected based on the alert rule and provides a monitoring result to the data transfer control 120. The sensor data collection control 140 collects data of operation sensors and temperature, humidity, or pressure sensors and provides them to the alert rule monitor 130.

(2-3) Various Data Next, the generation of various data handled by the operation management system and the contents of the various data will be described with reference to FIGS.

  As described above, the sensor data collection control 140 of the target site 100 collects data detected by the operation sensors 106 of various devices, the temperature sensor 102, the humidity sensor 103, the atmospheric pressure sensor 104, and the like.

  For example, when the device A 107 installed in the target base 100 is a turbo refrigerator, the sensor data collection control 140 collects the operation state, alarm state, power consumption, etc. of the device from the refrigeration device control panel sensor, Cooling water inlet temperature, cooling water outlet temperature, cooling water inlet temperature, cooling water outlet temperature are collected with a temperature sensor, heating degree is collected with a heating degree sensor in the evaporation coil, and supercooling degree is collected in the condensing coil. Or from the supercooling sensor. In addition, the power consumption of the compressor can be collected with a current sensor, the housing temperature of the compressor can be collected with a temperature sensor, the differential pressure of the compressor can be collected with a differential pressure sensor, the outdoor temperature can be collected with a temperature sensor, Humidity is collected with a humidity sensor, or outdoor atmospheric pressure is collected with a pressure sensor.

  The sensor data collection control 140 accumulates the collected sensor data as the sign analysis sensor data 142. Then, the sensor data collection control 140 generates learning sensor data 141 with a reduced collection frequency from the sign analysis sensor data 142 for a predetermined period.

  The sensor data collection control 140 provides learning sensor data 141 to the data transfer control 120. The sensor data collection control 140 provides the sign data for predictive analysis 142 to the alert rule monitor 130.

  The alert rule monitor 130 calculates the feature amount of the sensor data for predictive analysis by the feature amount calculation 132 using the feature amount calculation formula designated in the predictive detection preliminary alert rule provided from the central base 200, and detects the sign detection. Using the threshold specified in the preliminary alert rule, it is monitored whether the feature amount calculated by the threshold monitoring 131 exceeds the threshold. The feature amount calculation formula designated in the sign detection preliminary alert rule is a combination of simple four arithmetic operations. In the target base 100, the feature amount is calculated by simple four arithmetic operations, and the sign detection is executed depending on whether or not the feature amount exceeds the threshold value. In this way, it is possible to perform preliminary sign detection without applying a processing load such as causing the target base 100 to perform complicated calculations.

  When the alert rule monitor 130 determines that the feature amount exceeds the threshold, the alert rule monitor 130 notifies the data transfer control 120 of a sign detection preliminary alert. The data transfer control 120 notified of the sign detection preliminary alert causes the sensor data collection control 140 to provide the sensor data 142 for sign analysis.

  The learning sensor data 141 is sensor data that is collected at the central base 200 and is used for model construction for sign detection. As shown in FIG. 4, the item number column 1410, the monitoring item column 1411, and the summary column 1412 are used. And a collection frequency column 1413.

  The item number column 1410 stores data item numbers. The monitoring item column 1411 stores information on monitoring items of devices. The summary column 1412 stores a summary of monitoring items. The collection frequency column 1413 stores the collection frequency of monitoring items.

  In FIG. 4, for example, the item number “1” of the learning sensor data 141 indicates that the monitoring item is “operating state” of the device, “active” or “stopped” is monitored, and the collection frequency is 30 minutes. It turns out that it is once. In addition, the item number “2” of the learning sensor data 141 indicates that the monitoring item is “alarm state” of the device, whether the abnormality is “present” or “not present”, and the sampling frequency is once every 30 minutes. I know that there is.

  The predictor analysis sensor data 142 is data for analyzing the details of the abnormal predictor when the predictive detection preliminary alert is generated at each base and is collected at the central base 200. As shown in FIG. A column 1420, a monitoring item column 1421, a summary column 1422, and a collection frequency column 1423 are configured.

  The item number column 1420 stores data item numbers. The monitoring item column 1421 stores information on monitoring items of the device. The summary column 1422 stores the summary of the monitoring items. The collection frequency column 1423 stores the collection frequency of monitoring items.

  In FIG. 5, for example, the item number “1” of the predictor analysis sensor data 142 indicates that the monitoring item is “operating state” of the device, and is monitored whether “operating” or “stopped”, and the collection frequency is 30. It turns out that it is once per second. In addition, the item number “2” in the learning sensor data 141 indicates that the monitoring item is the “alarm state” of the device, whether the abnormality is “present” or “not present”, and the sampling frequency is once every 30 seconds. I know that there is.

  As described above, the learning sensor data 141 and the sign analysis sensor data 142 have the same items to be monitored, but the sign analysis sensor data 142 is data that is collected more frequently than the learning sensor data 141. .

  For example, when the size of data collected from each sensor is 3 bytes (3B), the data amount of learning sensor data of a certain sensor is 120 B / hour (3B × 20 × 2 times / hour), and predictive analysis The amount of sensor data is 7200 B / h (3B × 20 × 60 × 2 times / hour). Therefore, in the present embodiment, the network traffic can be reduced by setting the data amount of the learning sensor data transmitted in the normal state to be smaller than the predictive analysis sensor data 142.

  When an abnormality detection alert is notified from the sensor data collection control 140 to the data transfer control 120, the data transfer control 120 causes the sensor data collection control 140 to collect and provide sensor data 143 for abnormality analysis. Further, when the data transfer control 120 is requested to send the abnormality analysis sensor data 143 from the central base 200, the sensor data collection control 140 collects and provides the abnormality analysis sensor data 143.

  The abnormality analysis sensor data 143 is data for identifying the cause of the abnormality of the device when an abnormality detection alert occurs. As shown in FIG. 6, the item number column 1430, the monitoring item column 1431, and the summary column 1432, a collection frequency column 1433, and a collection time 1434.

  The item number column 1430 stores data item numbers. The monitoring item column 1431 stores information on monitoring items of the device. The summary column 1432 stores a summary of monitoring items. The collection frequency column 1433 stores the collection frequency of the monitoring item. In the collection time 1434, the collection time of the monitoring item is stored.

  In FIG. 6, for example, the item number “1” of the sensor data 143 for abnormality analysis indicates that the monitoring item is “operating state” of the device, and “in operation” or “stopped” is monitored. It can be seen that the sampling time is 10 minutes per second. In addition, for example, the item number “13” of the sensor data 143 for abnormality analysis is the monitoring item “vibration of the compressor 1” of the device, the “acceleration value of the vibration of the shaft of the compressor 1” is monitored, and the sampling frequency is It can be seen that the sampling time is 10 minutes per 1 ms. For example, when the motor of the compressor rotates at 3600 rpm (60 Hz) and there is an abnormality related to the rotation axis, vibration of an integral multiple (n) of 60 Hz occurs. In order to frequency-analyze the vibration in which such an abnormality occurs, the sampling frequency needs to be (60 × n × 2 × N) Hz. Here, N is an integer related to the frequency resolution, and is usually an integer smaller than 10. The item number 13 indicates the sampling frequency when n = 4 and N = 2, for example.

  For example, when the size of data collected from each sensor is 3 bytes (3B), the data amount of sensor data for abnormality analysis of a certain sensor is 36 MB / 10 minutes (3B × 18 × 2 times × 10 + 3B × 2 × 1000 × 60 × 10). As described above, the sensor data for abnormality analysis is data at a specific time including the time when the abnormality occurs, and data in which a valid data item is added to the sensor data for predictive analysis to identify the cause of the abnormality. It can be seen that it is.

  Returning to FIG. 3, the data transfer control 120 causes the transfer prioritization 121 to prioritize the data transfer for the various data and alerts provided. Then, the data transfer control 120 causes the transfer priority control 122 to transfer various data and alerts in accordance with the priority order according to the transfer priority ordering 121. Also, in the data transfer control 230 of the central site 200, the transfer priority order 231 is assigned a data transfer priority to various data and alerts, and is transferred by the transfer priority control 232 according to the priority order.

  The data transfer control 120 of the target site 100 and the data transfer control 230 of the central site 200 transfer data according to a common transfer priority when transmitting and receiving various data and alerts. For example, the abnormality detection alert, the sign detection preliminary alert, the policy information, the sign analysis sensor data, and the learning sensor data are arranged in the order of priority. As a result, information related to the abnormality of the device can be preferentially transferred.

(3) Operation Management Processing Next, operation management processing in the operation management system according to the present embodiment will be described with reference to FIGS. FIG. 7 shows operation management processing in a normal state where the device is operating normally, and FIG. 8 shows operation management processing in an abnormality detection state in which an abnormality of the device is detected. FIG. 9 and FIG. 10 show the operation management process when a sign detection preliminary alert occurs in the target site 100. FIG. 11 and FIG. 12 show the operation management process when it is determined that there is a possibility of the sign detection at the central site 200 (prediction detection preliminary determination).

(3-1) Operation Management Processing in Normal State As shown in FIG. 7 where the device is operating normally, the BMS 105 at the base 100 collects sensor data from the operation sensor 106 of each device 107. Based on the conditions determined by the design specifications of each device 107, the abnormality detection of each device 107 is monitored (S100).

  Then, the sensor data collection control 140 of the target site 100 collects sensor data from the BMS 105 and accumulates it as the sign analysis sensor data 142 (S101). Subsequently, the sensor data collection control 140 generates learning sensor data 141 whose collection frequency is reduced from the sensor data 142 for predictive analysis for a certain period (S102), and the data transfer control 120 stores the learning sensor data 141 in the central location. 200 (S103).

  The model construction (learning) 250 of the central base 200 constructs a model of each device by analyzing and learning the learning sensor data 141 transmitted from the target base 100, and generates a predictive detection preliminary alert rule based on the model. 260 generates a sign detection preliminary alert rule including a feature amount calculation formula and a threshold value for preliminarily detecting an abnormal sign (S104).

  Then, the data transfer control 230 of the central site 200 transmits a policy including the sign detection preliminary alert rule generated in step S104 to the corresponding site 100 (S105).

  The sensor data collection control 140 generates learning sensor data 141 whose collection frequency is reduced from the sign analysis sensor data 142 for a predetermined period (S106), and the data transfer control 120 stores the learning sensor data 141 in the central base 200. (S107).

  Then, the alert rule monitor 130 of the base 100 to which the sign detection preliminary alert rule is transmitted from the central base 200 in step S105, collects sensor data based on the feature amount calculation formula and the threshold value specified in the sign detection preliminary alert rule. Monitoring for the sign analysis sensor data 142 received from the control 140 is started (S108).

  Then, the model construction (learning) 250 of the central base 200 performs analysis and learning by adding the learning sensor data 141 transmitted in step S107 to the learning sensor data analyzed and learned in step S104, thereby achieving higher accuracy. The predictive detection preliminary alert rule generation 260 updates the predictive detection preliminary alert rule based on the model (S109).

  Then, the data transfer control 230 of the central site 200 transmits the policy including the sign detection preliminary alert rule updated in step S109 to the corresponding site 100 (S110).

  Further, the sensor data collection control 140 generates the learning sensor data 141 with the collection frequency reduced from the sign analysis sensor data 142 for a certain period (S111), and the data transfer control 120 stores the learning sensor data 141 in the central base 200. (S112).

  Then, the alert rule monitor 130 of the base 100 to which the updated sign detection preliminary alert rule is transmitted from the central base 200 in step S110 has the feature amount calculation formula and the threshold value specified in the updated sign detection preliminary alert rule. At the same time, monitoring of the sign analysis sensor data 142 received from the sensor data collection control 140 is started (S113).

  In normal times, steps S109 to S113 are repeated thereafter.

(3-2) Operation Management Process in Abnormality Detection State FIG. 8 is an explanatory diagram for explaining the operation management process in the abnormality detection state in which an abnormality of the device is detected. In FIG. 8, as in FIG. 7, the BMS 105 at the base 100 collects sensor data from the operation sensors 106 of the respective devices 107, and the abnormalities of the respective devices 107 are determined based on the conditions defined by the design specifications of the respective devices 107. Detection monitoring is performed (S200).

  And step S201-step S206 are implemented similarly to the repetition process of the above-mentioned normal state. That is, the sensor data collection control 140 of the target site 100 collects sensor data from the BMS 105 and accumulates it as the sign analysis sensor data 142 (S201). Subsequently, the sensor data collection control 140 generates learning sensor data 141 whose collection frequency is reduced from the sensor data 142 for predictive analysis for a certain period (S202), and the data transfer control 120 stores the learning sensor data 141 in the central location. 200 (S203).

  Then, the model construction (learning) 250 of the central base 200 analyzes and learns by adding the learning sensor data 141 transmitted in step S203 to the learning sensor data that has been analyzed and learned earlier, thereby obtaining higher accuracy. The device model is constructed, and the sign detection preliminary alert rule generation 260 updates the sign detection preliminary alert rule based on the model (S204).

  Then, the data transfer control 230 of the central site 200 transmits the policy including the sign detection preliminary alert rule updated in step S109 to the corresponding site 100 (S205).

  Then, the alert rule monitor 130 of the base 100 to which the updated sign detection preliminary alert rule is transmitted from the central base 200 in step S205 has the feature amount calculation formula and the threshold value specified in the updated sign detection preliminary alert rule. At the same time, monitoring of the sign analysis sensor data 142 received from the sensor data collection control 140 is started (S206).

  When the BMS 105 detects an abnormality of the apparatus, the data transfer control 120 of the target base 100 transmits an abnormality detection alert to the central base 200 (S207).

  Upon receipt of the abnormality detection alert in step S207, the policy generation 240 of the central site 200 shifts to the abnormality detection state and generates a policy requesting the abnormality analysis sensor data 143 (S208). Then, the data transfer control 230 transmits the policy generated in step S208 to the corresponding base 100 (S209).

  After the abnormality detection alert is transmitted to the central site 200 in step S207, if the predictive detection preliminary alert rule is not transmitted from the central site 200, the target site 100 transmits from the central site 200 before the abnormality detection alert is generated. Based on the predicted sign detection preliminary alert rule, the sensor data 142 for sign analysis is monitored (S217). In addition, in step S217, when a predictive detection preliminary alert rule generated using data of another base is transmitted from the central base 200, a predictive analysis sensor based on the predictive detection preliminary alert rule Data 142 may be monitored.

  The sensor data collection control 140 of the target base 100 collects the abnormality analysis sensor data 143 (S210), and the data transfer control 120 transmits the abnormality analysis sensor data 143 to the central base 200 (S211).

  The abnormality cause analysis 210 of the central base 200 analyzes the abnormality analysis sensor data 143 transmitted in step S211 and identifies the cause of the abnormality (S212). Then, an abnormality countermeasure is notified to the maintenance staff at the target site (S213).

  When a notification of completion of abnormality countermeasures is received from the maintenance staff at the target site (S214), the state is recovered from the abnormality detection state, and the policy generation 240 of the central base 200 generates a policy notifying the recovery of the abnormality detection state (S215). Then, the policy generated by the data transfer control 230 is transmitted to the base 100 (S216).

  When the site 100 receives the policy in step S216, the abnormality detection state is recovered, the normal state is restored, and steps S201 to S206 are repeated thereafter.

(3-3) Operation Management Process When Prediction Detection Preliminary Alert Occurs FIGS. 9 and 10 are explanatory diagrams for explaining the operation management process when a sign detection preliminary alert occurs. FIG. 9 shows sensor data for predictive analysis. FIG. 10 shows a case where it is determined that there is an abnormal sign as a result of analyzing the sensor data for sign analysis.

  As shown in FIG. 9, the BMS 105 of the base 100 collects sensor data from the operation sensor 106 of each device 107 and detects abnormality of each device 107 based on the conditions defined by the design specifications of each device 107. Monitoring is performed (S300).

  And step S301-step S306 are implemented similarly to the repetition process of the above-mentioned normal state. That is, the sensor data collection control 140 of the target site 100 collects sensor data from the BMS 105 and accumulates it as the sign analysis sensor data 142 (S301). Subsequently, the sensor data collection control 140 generates learning sensor data 141 whose collection frequency is reduced from the sensor data 142 for predictive analysis for a certain period (S302), and the data transfer control 120 stores the learning sensor data 141 in the central location. 200 (S303).

  Then, the model construction (learning) 250 of the central base 200 analyzes and learns by adding the learning sensor data 141 transmitted in step S303 to the learning sensor data that has been analyzed and learned earlier, thereby obtaining higher accuracy. A model of the device is constructed, and the sign detection preliminary alert rule generation 260 updates the sign detection preliminary alert rule based on the model (S304).

  Then, the data transfer control 230 of the central base 200 transmits the policy including the sign detection preliminary alert rule updated in step S304 to the base 100 (S305).

  Then, the alert rule monitor 130 of the base 100 to which the updated sign detection preliminary alert rule is transmitted from the central base 200 in step S305 has the feature amount calculation formula and the threshold value specified in the updated sign detection preliminary alert rule. At the same time, monitoring of the sign analysis sensor data 142 received from the sensor data collection control 140 is started (S306).

  When the alert rule monitor 130 of the target site 100 monitors the sign analysis sensor data 142 based on the feature amount calculation formula and the threshold value of the sign detection preliminary alert rule, it is determined that there is a possibility of sign detection. The data transfer control 120 transmits a sign detection preliminary alert to the central site 200 (S307).

  The policy generation 240 of the central base 200 that has received the sign detection preliminary alert in step S307 shifts to the sign detection preliminary state, and generates a policy for requesting the sign analysis sensor data 142 (S308). Then, the data transfer control 230 transmits the policy generated in step S308 to the corresponding base 100 (S309).

  If the sign detection preliminary alert rule is not transmitted from the central base 200 after the sign detection preliminary alert is transmitted to the central base 200 in step S307, the target base 100 receives from the central base 200 before the abnormality detection alert is generated. Based on the transmitted sign detection preliminary alert rule, the sign analysis sensor data 142 is monitored (S315). In addition, in step S315, when a predictive detection preliminary alert rule generated using data of other bases is transmitted from the central base 200, a predictive analysis sensor based on the predictive detection preliminary alert rule Data 142 may be monitored.

  The sensor data collection control 140 of the target base 100 collects the sign analysis sensor data for a predetermined period (S310), and the data transfer control 120 transmits the sign analysis sensor data 142 to the central base 200 (S311). .

  The detailed sign analysis 220 of the central base 200 analyzes the sign analysis sensor data 142 transmitted in step S311 and determines whether there is an abnormal sign (S312). If it is determined by the predictive detail analysis 220 that there is no abnormal sign, the sign detection standby state is recovered, and the policy generation 240 generates a policy notifying the recovery of the sign detection preliminary state (S313), and the data transfer control 230 generates The policy is transmitted to the base 100 (S314).

  When the site 100 receives the policy in step S314, the sign detection preliminary state is restored, and the normal state is restored. Thereafter, steps S301 to S306 are repeated.

  As shown in FIG. 10, the BMS 105 of the base 100 collects sensor data from the operation sensor 106 of each device 107 and detects abnormality of each device 107 based on the conditions defined by the design specifications of each device 107. Monitoring is performed (S400).

  And step S401-step S406 are implemented similarly to the repetition process of the above-mentioned normal state. That is, the sensor data collection control 140 of the target base 100 collects sensor data from the BMS 105 and accumulates it as the sign analysis sensor data 142 (S401). Subsequently, the sensor data collection control 140 generates learning sensor data 141 whose collection frequency is reduced from the sign analysis sensor data 142 for a predetermined period (S402), and the data transfer control 120 stores the learning sensor data 141 in the central location. 200 (S403).

  Then, the model building (learning) 250 of the central base 200 analyzes and learns by adding the learning sensor data 141 transmitted in step S403 to the learning sensor data analyzed and learned earlier, thereby obtaining higher accuracy. A model of the device is constructed, and the sign detection preliminary alert rule generation 260 updates the sign detection preliminary alert rule based on the model (S404).

  Then, the data transfer control 230 of the central site 200 transmits the policy including the sign detection preliminary alert rule updated in step S404 to the corresponding site 100 (S405).

  Then, the alert rule monitor 130 of the base 100 to which the updated sign detection preliminary alert rule is transmitted from the central base 200 in step S405 has the feature amount calculation formula and the threshold value specified in the updated sign detection preliminary alert rule. At the same time, monitoring of the sign analysis sensor data 142 received from the sensor data collection control 140 is started (S406).

  When the alert rule monitor 130 of the target site 100 monitors the sign analysis sensor data 142 based on the feature amount calculation formula and the threshold value of the sign detection preliminary alert rule, it is determined that there is a possibility of sign detection. The data transfer control 120 transmits a sign detection preliminary alert to the central site 200 (S407).

  The policy generation 240 of the central base 200 that has received the sign detection preliminary alert in step S407 shifts to the sign detection preliminary state, and generates a policy for requesting the sign analysis sensor data 142 (S408). Then, the data transfer control 230 transmits the policy generated in step S408 to the base 100 (S409).

  If the sign detection preliminary alert rule is not transmitted from the central base 200 after the sign detection preliminary alert is transmitted to the central base 200 in step S407, the target base 100 receives the warning from the central base 200 before the abnormality detection alert is generated. Based on the transmitted sign detection preliminary alert rule, the sign analysis sensor data 142 is monitored (S422). In addition, in step S422, when the sign detection preliminary alert rule generated using the data of another base is transmitted from the central base 200, the sign analysis sensor based on the sign detection preliminary alert rule is transmitted. Data 142 may be monitored.

  The sensor data collection control 140 of the target base 100 collects the sign analysis sensor data for a predetermined period (S410), and the data transfer control 120 transmits the sign analysis sensor data 142 to the central base 200 (S411). .

  The detailed sign analysis 220 of the central base 200 analyzes the sign analysis sensor data 142 transmitted in step S411, and determines whether there is an abnormal sign (S412). When it is determined by the predictive detail analysis 220 that there is an abnormal sign, the state shifts to the abnormal sign state, and the policy generation 240 generates a policy requesting the abnormality analysis sensor data 143 (S413). Then, the data transfer control 230 transmits the policy generated in step S413 to the corresponding base 100 (S414).

  The sensor data collection control 140 of the target base 100 collects the abnormality analysis sensor data 143 (S415), and the data transfer control 120 transmits the abnormality analysis sensor data 143 to the central base 200 (S416).

  The abnormality cause analysis 210 of the central base 200 analyzes the abnormality analysis sensor data 143 transmitted in step S416 to identify the cause of the abnormality (S417). Then, an abnormality sign countermeasure is notified to the maintenance staff at the target site (S418).

  When a notice of completion of the abnormality sign countermeasure is received from the maintenance staff at the target base (S419), the state is recovered from the abnormal sign state, and the policy generation 240 of the central base 200 generates a policy notifying the recovery of the abnormal sign state (S420). ), The policy generated by the data transfer control 230 is transmitted to the base 100 (S421).

  When the site 100 receives the policy in step S421, the abnormal sign state is recovered and the normal state is restored, and thereafter steps S401 to S406 are repeated.

(3-4) Operation Management Process at Prediction Detection Preliminary Judgment FIGS. 11 and 12 explain the operation management process when it is determined that there is a possibility of the sign detection at the central base 200 (prediction detection preliminary determination). FIG. 11 shows a case where it is determined that there is no abnormal sign as a result of analyzing the sensor data for predictive analysis, and FIG. 12 shows a case where it is determined that there is an abnormal sign as a result of analyzing the sensor data for predictive analysis. Show.

  As shown in FIG. 11, the BMS 105 of the base 100 collects sensor data from the operation sensor 106 of each device 107, and detects abnormality of each device 107 based on the conditions defined by the design specifications of each device 107. Monitoring is performed (S500).

  And step S501-step S506 are implemented similarly to the repetition process of the above-mentioned normal state. That is, the sensor data collection control 140 of the target site 100 collects sensor data from the BMS 105 and accumulates it as the sign analysis sensor data 142 (S501). Subsequently, the sensor data collection control 140 generates learning sensor data 141 whose collection frequency has been reduced from the sign analysis sensor data 142 for a predetermined period (S502), and the data transfer control 120 stores the learning sensor data 141 in the central location. 200 (S503).

  Then, the model construction (learning) 250 of the central base 200 analyzes and learns by adding the learning sensor data 141 transmitted in step S503 to the learning sensor data 141 that has been analyzed and learned earlier, thereby achieving higher accuracy. The predictive detection preliminary alert rule generation 260 updates the predictive detection preliminary alert rule based on the model (S504).

  Then, the data transfer control 230 of the central site 200 transmits the policy including the sign detection preliminary alert rule updated in step S504 to the corresponding site 100 (S505).

  Then, the alert rule monitor 130 of the base 100 to which the updated sign detection preliminary alert rule is transmitted from the central base 200 in step S505 has the feature amount calculation formula and the threshold value specified in the updated sign detection preliminary alert rule. At the same time, monitoring of the sign analysis sensor data 142 received from the sensor data collection control 140 is started (S506).

  Then, when the detailed sign analysis 220 of the central base 200 analyzes the learning sensor data 141 and determines that there is a possibility that an abnormal sign is detected (S507), the process proceeds to the sign detection preliminary state to generate a policy. 240 generates a policy for requesting the sensor data 142 for predictive analysis (S508). Then, the data transfer control 230 transmits the policy generated in step S508 to the corresponding base 100 (S509).

  If it is determined in step S507 that there is a possibility that an anomaly sign is detected, if the predictive detection preliminary alert rule is not transmitted from the central office 200, the target base 100 will start from the central office 200 before the anomaly detection alert is generated. Based on the transmitted sign detection preliminary alert rule, the sign analysis sensor data 142 is monitored (S515). In addition, in step S515, when a predictive detection preliminary alert rule generated using data of another base is transmitted from the central base 200, a predictive analysis sensor based on the predictive detection preliminary alert rule Data 142 may be monitored.

  The sensor data collection control 140 of the target base 100 collects the sign analysis sensor data for a predetermined period (S510), and the data transfer control 120 transmits the sign analysis sensor data 142 to the central base 200 (S511). .

  The detailed indication analysis 220 of the central base 200 analyzes the indication analysis sensor data 142 transmitted in step S511, and determines whether there is an abnormal indication (S512). When it is determined by the predictive detail analysis 220 that there is no abnormal sign, the sign detection preliminary state is restored, and the policy generation 240 generates a policy notifying the recovery of the sign detection preliminary state (S513), and the data transfer control 230 generates The policy is transmitted to the base 100 (S514).

  When the site 100 receives the policy in step S514, the sign detection preliminary state is restored, and the normal state is restored. Thereafter, steps S501 to S506 are repeated.

  As shown in FIG. 12, the BMS 105 of the base 100 collects sensor data from the operation sensor 106 of each device 107, and detects abnormality of each device 107 based on the conditions defined by the design specifications of each device 107. Monitoring is performed (S600).

  And step S601-step S606 are implemented similarly to the above-mentioned repetition process of a normal state. That is, the sensor data collection control 140 of the target site 100 collects sensor data from the BMS 105 and accumulates it as the sign analysis sensor data 142 (S601). Subsequently, the sensor data collection control 140 generates learning sensor data 141 whose collection frequency is reduced from the sensor data 142 for predictive analysis for a certain period (S602), and the data transfer control 120 stores the learning sensor data 141 in the central location. 200 (S603).

  Then, the model construction (learning) 250 of the central base 200 analyzes and learns by adding the learning sensor data 141 transmitted in step S603 to the learning sensor data 141 that has been analyzed and learned earlier, thereby achieving higher accuracy. The predictive detection preliminary alert rule generation 260 updates the predictive detection preliminary alert rule based on the model (S604).

  Then, the data transfer control 230 of the central site 200 transmits the policy including the sign detection preliminary alert rule updated in step S604 to the corresponding site 100 (S605).

  Then, the alert rule monitor 130 of the base 100 to which the updated sign detection preliminary alert rule is transmitted from the central base 200 in step S110 has the feature amount calculation formula and the threshold value specified in the updated sign detection preliminary alert rule. At the same time, monitoring of the sign analysis sensor data 142 received from the sensor data collection control 140 is started (S606).

  Then, when the detailed sign analysis 220 of the central base 200 analyzes the learning sensor data 141 and determines that there is a possibility of detection of an abnormal sign (S607), the process proceeds to the sign detection preliminary state to generate a policy. 240 generates a policy requesting the sign analysis sensor data 142 (S608). Then, the data transfer control 230 transmits the policy generated in step S608 to the corresponding base 100 (S609).

  If it is determined in step S607 that there is a possibility that an anomaly sign has been detected, if the predictive detection preliminary alert rule is not transmitted from the central office 200, the target base 100 will start from the central office 200 before the anomaly detection alert is generated. Based on the transmitted sign detection preliminary alert rule, the sign analysis sensor data 142 is monitored (S622). In addition, in step S622, when the sign detection preliminary alert rule generated using the data of other bases is transmitted from the central base 200, the sign analysis sensor based on the sign detection preliminary alert rule is transmitted. Data 142 may be monitored.

  The sensor data collection control 140 at the target site 100 collects sensor data for predictive analysis during a predetermined period (S610), and the data transfer control 120 transmits the sensor data 142 for predictive analysis to the central site 200 (S611). .

  The detailed sign analysis 220 of the central base 200 analyzes the sign analysis sensor data 142 transmitted in step S611, and determines whether there is an abnormal sign (S612). When it is determined by the predictive detail analysis 220 that there is an abnormal sign, the state shifts to an abnormal sign state, and the policy generation 240 generates a policy requesting the sensor data 143 for abnormality analysis (S613). Then, the data transfer control 230 transmits the policy generated in step S613 to the base 100 (S614).

  The sensor data collection control 140 of the target site 100 collects the abnormality analysis sensor data 143 (S615), and the data transfer control 120 transmits the abnormality analysis sensor data 143 to the central site 200 (S616).

  The abnormality cause analysis 210 of the central base 200 analyzes the abnormality analysis sensor data 143 transmitted in step S616, and analyzes the cause of the abnormality (S617). Then, an abnormality sign countermeasure is notified to the maintenance staff at the target site (S618).

  When a notice of completion of the abnormality sign countermeasure is received from the maintenance staff at the target base (S619), recovery from the abnormal sign state is performed, and the policy generation 240 of the central base 200 generates a policy notifying the recovery of the abnormal sign state (S620). ), And transmits the policy generated by the data transfer control 230 to the base 100 (S621).

  When the site 100 receives the policy in step S621, the abnormal sign state is recovered and the normal state is restored, and thereafter steps S601 to S606 are repeated.

(4) Effect of this Embodiment According to the above embodiment, the control unit 101 of the target site 100 detects an apparatus abnormality from the operation data based on a predetermined condition defined by the design specifications of the device 107. An anomaly detection alert is generated, and an anomaly detection is detected from the operation data based on an anomaly detection preliminary alert rule for preliminarily detecting an anomaly sign designated at the central base 200. The control unit of the central base 200 constructs a model of the device for detecting a sign for detecting a sign of abnormality based on the operation data of the devices 107 collected from the plurality of target bases 100, Based on the constructed device model, a sign detection preliminary alert rule is generated, and the constructed device model and the operation data of the target device 107 at the target base 100 are utilized. Then, the sign analysis for analyzing the sign of abnormality is executed, and the cause of abnormality is analyzed based on the operation data of the target device 107 at the target base 100.

  As a result, when the operation data of a large number of target sites is aggregated at the central site and an abnormality sign is detected, a preliminary detection detection judgment for detecting an abnormality sign from the operation data is performed at each site, which is part of the detection process. Thus, by generating the sign detection preliminary alert, it is possible to suppress a reduction in the amount of data transfer and an increase in computing resources at the central base.

DESCRIPTION OF SYMBOLS 100 Target site 101 Control part 102 Temperature sensor 103 Humidity sensor 104 Atmospheric pressure sensor 106 Operation sensor 107 Equipment 110 Policy base control 120 Data transfer control 122 Transfer priority control 130 Alert rule monitoring 131 Threshold monitoring 132 Feature amount calculation 140 Sensor data collection control 200 Central Base 210 Abnormal cause analysis 220 Predictive detail analysis 230 Data transfer control 232 Transfer priority control 240 Policy generation 260 Predictive detection preliminary alert rule generation

Claims (9)

An operation management system that supports operation management by monitoring the operation status of devices installed at multiple target sites at a central site,
The plurality of target bases are:
An operation sensor for collecting operation data for monitoring the operation status of the device;
A storage unit for storing operation data collected by the operation sensor;
Determine the collection item and collection frequency of the operation data, and transfer the operation data to the central base,
With
The central base is
A storage unit for storing the operation data;
A control unit for determining an operating state of an apparatus installed at the plurality of target sites based on the operating data;
With
The control unit of the target site is
Based on a predetermined condition defined in the design specification of the device, an abnormality of the device is detected from the operation data to generate an abnormality detection alert,
Based on the sign detection preliminary alert rule for preliminarily detecting the sign of abnormality designated at the central base, detecting the sign of abnormality from the operation data and generating a sign detection preliminary alert,
The control unit of the central base is
Build a model of the device for sign detection for detecting a sign of abnormality based on operation data collected from the plurality of target sites,
Based on the model of the constructed device, the predictive detection preliminary alert rule is generated,
Perform predictive analysis that analyzes the predictor of abnormality by utilizing the operation data of the target device at the target base and the model of the constructed device,
Perform an abnormality cause analysis to analyze the cause of the abnormality based on the operation data of the target device of the target base,
The control unit of the target site is
The operation data is
As sensor data for learning for model construction for the sign detection,
Sensor data for predictive analysis for the predictive analysis,
As sensor data for abnormal cause analysis for the abnormal cause analysis,
Store in the storage unit, transfer in response to a request from the central base,
Collecting the learning sensor data and the sign analysis sensor data in the same collection item,
Storing the learning sensor data recording frequency lower than the sign analysis sensor data recording frequency in the storage unit;
Transfer the learning sensor data to the central base during normal operation of the device,
The warning detection preliminary alert, characterized in that for transferring the sensor data for the sign analysis to the central hub when the caused, the management system for luck. The control unit of the target site is
As the abnormal cause analysis sensor data, add a collection item effective for identifying the cause of abnormality of the device to the collection item of the learning sensor data and store it in the storage unit,
The operation management system according to claim 1 , wherein when the abnormality detection alert is generated, the abnormality cause analysis sensor data is transferred to the central base. The control unit of the target site is
As the abnormal cause analysis sensor data,
The operation management system according to claim 2 , wherein data of a vibration sensor that analyzes a frequency is collected in order to analyze an abnormality cause of the device. The control unit of the central base is
Based on the model of the constructed device, generate a feature amount calculation formula for calculating a predetermined feature amount and the predictive detection preliminary alert rule including a predetermined threshold value of the feature amount,
The control unit of the target site is
A feature amount of the sensor data for predictive analysis is calculated by the feature amount calculation formula included in the predictive detection preliminary alert rule, and the predictive detection preliminary alert is generated when the feature amount exceeds the threshold value. The operation management system according to claim 3 . The control unit of the target site and the control unit of the central site,
The abnormality detection alert, the abnormality cause analysis sensor data, the warning detection preliminary alert, the sign analysis sensor data, the warning detection preliminary alert rules, and characterized in that data is transferred in the order of the learning sensor data, The operation management system according to claim 4 . The control unit of the target site is
Adding external environment data including temperature, humidity, and atmospheric pressure at the target base to the operation data to obtain the sensor data for learning.
The operation management system according to claim 5 , wherein: The control unit of the central base is
Utilizing the external environment data including the temperature, humidity, and pressure at the target base that the control unit of the target base added to the sensor data for learning,
7. The operation according to claim 6 , wherein a single model is constructed for all devices having the same design specifications installed at target sites distributed in a wide area and located in various climate regions. Management system. The apparatus installed in the target base for heating, ventilation and air conditioning equipment such as air conditioning: characterized in that it is a (HVAC heating ventilating air conditioning), the operation management system according to any of claims 1 to 7 . An operation management method in an operation management system that supports operation management by monitoring the operation status of devices installed at a plurality of target sites in a central site,
The control unit of the target site determines a collection item and collection frequency of operation data for monitoring the operation status of the device, and transfers the operation data to the central site.
A step of constructing a model of the device for sign detection for detecting a sign of abnormality based on operation data collected from the plurality of target sites by the control unit of the central site;
The control unit of the central base generates a sign detection preliminary alert rule for preliminarily detecting a sign of abnormality based on the model of the constructed device;
The control unit of the target base detects a sign of abnormality from the operation data based on the sign detection preliminary alert rule specified at the central base, and generates a sign detection preliminary alert;
When the predictive detection preliminary alert is generated, the central base control unit performs predictive analysis to analyze the abnormal sign by utilizing the model of the constructed device and the operation data of the target device of the target base. Steps to perform;
The control unit of the target base detects an abnormality of the apparatus from the operation data and generates an abnormality detection alert based on a predetermined condition defined in the design specification of the apparatus;
When the abnormality detection alert is generated, or when it is determined that there is an abnormal sign as a result of the predictive analysis, the control unit of the central base is abnormal based on the operation data of the target device of the target base Performing an abnormal cause analysis to analyze the cause of
Only including,
The control unit of the target site is
The operation data is
As sensor data for learning for model construction for the sign detection,
Sensor data for predictive analysis for the predictive analysis,
Sensor data for abnormality cause analysis for the abnormality cause analysis,
The learning sensor data and the sign analysis sensor data are collected by the same collection item, and the recording frequency of the learning sensor data is lowered from the recording frequency of the sign analysis sensor data, and is collected by the operation sensor. Stored in the storage unit that stores the operation data,
In the transferring step,
Transfer the learning sensor data to the central base during normal operation of the device,
The operation management method according to claim 1, wherein when the sign detection preliminary alert is generated, the sign analysis sensor data is transferred to the central base .

Images