Maintenance work support device, maintenance work support method, and maintenance work support program

Abstract

Provided are a maintenance work assistance device, a maintenance work assistance method, and a maintenance work assistance program, which can assist maintenance work by reducing the number of times of operation by a maintenance worker. The maintenance work assistance device includes: a first acquisition unit that acquires a data set including a combination of device information of a target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device; a second acquisition unit that acquires work content information in which a replaced or repaired component or a new replaced component, which is the content of a maintenance work performed by a maintenance worker with respect to the target device, is recorded; and a learning unit that learns so as to associate the data set acquired by the first acquisition unit with the replaced or repaired component or the replaced new component recorded in the job content information acquired by the second acquisition unit.

Description

[0001] This application is a divisional application of a patent application for "Maintenance Work Assistance Device, Maintenance Work Assistance Method, and Maintenance Work Assistance Program", with an application date of March 9, 2020, and an application number of 202080010489.4 (international application number PCT / JP2020 / 010024). Technical Field

[0002] The present disclosure relates to a maintenance work assistance device, a maintenance work assistance method, and a maintenance work assistance program. Background Art

[0003] In the prior art, the following technologies have been proposed, that is, diagnostic technologies for collecting in advance operation information of air-conditioning equipment during operation and diagnosing the cause of an abnormality in the event of an abnormality, and prediction technologies for predicting the location of an abnormality based on an abnormality code output by the air-conditioning equipment. According to these technologies, when an abnormality occurs, maintenance workers can perform inspections based on the diagnosed cause of the abnormality and the predicted location of the abnormality, and thus maintenance work such as determining a faulty component, replacing or repairing the faulty component can be efficiently performed.

[0004] [Cited Document]

[0005] [Patent Document]

[0006] [Patent Document 1] Japanese Patent Application Laid-Open No. 2004-265159 Summary of the Invention

[0007] [Technical Problem to be Solved]

[0008] On the other hand, when an abnormality occurs, even if the cause of the abnormality and the location of the abnormality are known in advance, maintenance workers cannot determine the faulty component unless they are present at the site. Therefore, the conventional diagnostic technologies and prediction technologies cannot reduce the number of times maintenance workers are dispatched.

[0009] The present disclosure provides a maintenance work assistance device, a maintenance work assistance method, and a maintenance work assistance program that can assist maintenance work by reducing the number of times maintenance workers are dispatched.

[0010] [Technical Solution]

[0011] The maintenance work assistance device according to the first aspect of the present disclosure includes:

[0012] a first acquisition unit that acquires a data set including a combination of device information of a target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device;

[0013] A second acquisition unit that acquires operation content information in which the content of the maintenance operation performed by the maintenance worker on the target device, that is, the replaced or repaired component or the new component after replacement, is recorded; and

[0014] A learning unit that performs learning by associating the data group acquired by the first acquisition unit with the replaced or repaired component or the new component after replacement recorded in the operation content information acquired by the second acquisition unit.

[0015] According to the first aspect of the present disclosure, the components to be replaced or repaired in the target device or the new components after replacement can be determined before dispatch, and the preparations required for replacement or repair can be made before dispatch, so the number of dispatches of the maintenance workers can be reduced. In other words, according to the first aspect of the present disclosure, a maintenance operation assistance device that can assist the maintenance operation by reducing the number of dispatches of the maintenance workers can be provided.

[0016] In addition, the maintenance operation assistance device according to the second aspect of the present disclosure has:[[]]

[0017] A first acquisition unit that acquires a data group including a combination of device information of the target device and operation information of the target device, or a data group including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device;

[0018] A second acquisition unit that acquires operation content information in which the maintenance operation performed by the maintenance worker on the target device, that is, the operation process, is recorded; and

[0019] A learning unit that performs learning by associating the data group acquired by the first acquisition unit with the operation process recorded in the operation content information acquired by the second acquisition unit.

[0020] According to the second aspect of the present disclosure, the operation process of the maintenance operation for the target device can be determined before dispatch, and the preparations required for executing the operation process can be made before dispatch, so the number of dispatches of the maintenance workers can be reduced. In other words, according to the second aspect of the present disclosure, a maintenance operation assistance device that can assist the maintenance operation by reducing the number of dispatches of the maintenance workers can be provided.

[0021] Furthermore, the maintenance operation assistance device according to the third aspect of the present disclosure has:[[]]

[0022] A calculation unit that calculates a reward based on evaluation information, which is information for evaluating the operation result of the maintenance operation for the target device; and

[0023] The learning unit learns about the components that should be replaced or repaired in the target device, or the new components after replacement, based on a data set that combines the device information of the target device and the operating information of the target device, or a data set that combines the device information of the target device and the phenomenon information indicating the phenomenon related to the target device.

[0024] Among them, the learning unit learns about the components that should be replaced or repaired in the target device, or the new components after replacement, based on the reward.

[0025] According to the third aspect of the present disclosure, the components that should be replaced or repaired in the target device, or the new components after replacement, can be determined before departure, and the preparations required for replacement or repair can be made before departure, so the number of departures of maintenance personnel can be reduced. In other words, according to the third aspect of the present disclosure, a maintenance operation assistance device that can assist maintenance operations by reducing the number of departures of maintenance personnel can be provided.

[0026] The maintenance operation assistance device according to the fourth aspect of the present disclosure has:

[0027] A calculation unit that calculates a reward based on evaluation information, which is information for evaluating the operation result of the maintenance operation for the target device; and

[0028] A learning unit that learns about the operation process of the maintenance operation for the target device based on a data set that combines the device information of the target device and the operating information of the target device, or a data set that combines the device information of the target device and the phenomenon information indicating the phenomenon related to the target device.

[0029] The learning unit learns about the operation process of the maintenance operation for the target device based on the reward.

[0030] According to the fourth aspect of the present disclosure, the operation process of the maintenance operation for the target device can be determined before departure, and the preparations required for executing the operation process can be made before departure, so the number of departures of maintenance personnel can be reduced. In other words, according to the fourth aspect of the present disclosure, a maintenance operation assistance device that can assist maintenance operations by reducing the number of departures of maintenance personnel can be provided.

[0031] In addition, the fifth aspect of the present disclosure is the maintenance operation assistance device according to any one of the first to fourth aspects, wherein the device information includes the category of the target device.

[0032] Furthermore, the sixth aspect of the present disclosure is the maintenance operation assistance device according to any one of the first to fourth aspects, wherein:

[0033] The operation information includes any one of the abnormal code, abnormal prediction data, operating conditions, indoor temperature, outdoor temperature, and cumulative working hours output by the target device.

[0034] In addition, the seventh aspect of the present disclosure is the maintenance operation assistance device according to the sixth aspect, wherein:

[0035] The target device is an air conditioning device, and the operating conditions include any one of the compressor speed, suction superheat degree, subcooling degree, exhaust temperature, high and low pressure values, high-pressure side temperature, low-pressure side temperature, valve opening degree, and external heat exchange temperature.

[0036] In addition, the eighth aspect of the present disclosure is the maintenance operation assistance device according to any one of the first to fourth aspects, wherein,

[0037] The phenomenon information is information indicating a phenomenon occurring due to a failure and / or abnormality of the target device.

[0038] In addition, the ninth aspect of the present disclosure is the maintenance operation assistance device according to the first aspect, and further has:

[0039] An inference unit that re-acquires a data group including a combination of device information of the target device and operation information of the target device, or a data group including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device, and infers, based on the learning result of the learning unit, a component to be replaced or repaired in the target device or a new component after replacement from the re-acquired data group.

[0040] In addition, the tenth aspect of the present disclosure is the maintenance operation assistance device according to the second aspect, and further has:

[0041] An inference unit that re-acquires a data group including a combination of device information of the target device and operation information of the target device, or a data group including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device, and infers, based on the learning result of the learning unit, an operation process of the maintenance operation for the target device from the re-acquired data group.

[0042] In addition, the maintenance operation assistance method according to the eleventh aspect of the present disclosure has:

[0043] A first acquisition step of acquiring a data group including a combination of device information of the target device and operation information of the target device, or a data group including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device;

[0044] The second acquisition step of obtaining operation content information, which records the content of the maintenance operation performed by the maintenance operator on the target device, that is, the replaced or repaired parts or the new parts after replacement; and

[0045] A learning step of performing learning by associating the data group obtained in the first acquisition step with the replaced or repaired parts or the new parts after replacement recorded in the operation content information obtained in the second acquisition step.

[0046] According to the eleventh aspect of the present disclosure, the parts to be replaced or repaired or the new parts after replacement in the target device can be determined before departure, and the preparations required for replacement or repair can be made before departure, so the number of departures of the maintenance operators can be reduced. In other words, according to the eleventh aspect of the present disclosure, a maintenance operation assistance method capable of assisting the maintenance operation by reducing the number of departures of the maintenance operators can be provided.

[0047] In addition, the maintenance operation assistance method according to the twelfth aspect of the present disclosure has:

[0048] The first acquisition step of obtaining a data group including a combination of device information of the target device and operation information of the target device, or a data group including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device;

[0049] The second acquisition step of obtaining operation content information, which records the maintenance operation performed by the maintenance operator on the target device, that is, the operation process; and

[0050] A learning step of performing learning by associating the data group obtained in the first acquisition step with the operation process recorded in the operation content information obtained in the second acquisition step.

[0051] According to the twelfth aspect of the present disclosure, the operation process of the maintenance operation for the target device can be determined before departure, and the preparations required for executing the operation process can be made before departure, so the number of departures of the maintenance operators can be reduced. In other words, a maintenance operation assistance method capable of assisting the maintenance operation by reducing the number of departures of the maintenance operators can be provided.

[0052] In addition, the maintenance operation assistance method according to the thirteenth aspect of the present disclosure has:

[0053] A calculation step of calculating a reward according to evaluation information, which is information for evaluating the operation result of the maintenance operation for the target device; and

[0054] Learning step: learning the components to be replaced or repaired in the target device or the new components after replacement based on a data group including a combination of device information of the target device and operation information of the target device, or based on a data group including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device.

[0055] Among them, the learning step learns the components to be replaced or repaired in the target device or the new components after replacement according to the reward.

[0056] According to the 13th aspect of the present disclosure, before dispatch, the components to be replaced or repaired in the target device or the new components after replacement can be determined, and the preparations required for replacement or repair can be made before dispatch, so the number of dispatches of maintenance personnel can be reduced. In other words, a maintenance operation assistance method that can assist maintenance operations by reducing the number of dispatches of maintenance personnel can be provided.

[0057] In addition, the maintenance operation assistance method according to the 14th aspect of the present disclosure has:

[0058] Calculation step: calculating a reward according to evaluation information, where the evaluation information is information for evaluating the operation result of a maintenance operation for a target device; and

[0059] Learning step: learning the operation process of a maintenance operation for the target device based on a data group including a combination of device information of the target device and operation information of the target device, or based on a data group including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device.

[0060] Among them, the learning step learns the operation process of a maintenance operation for the target device according to the reward.

[0061] According to the 14th aspect of the present disclosure, before dispatch, the operation process of a maintenance operation for the target device can be determined, and the preparations required for executing the operation process can be made before dispatch, so the number of dispatches of maintenance personnel can be reduced. In other words, a maintenance operation assistance method that can assist maintenance operations by reducing the number of dispatches of maintenance personnel can be provided.

[0062] In addition, the maintenance operation assistance program according to the 15th aspect of the present disclosure can cause a computer to execute the following steps, that is:

[0063] First acquisition step: acquiring a data group including a combination of device information of the target device and operation information of the target device, or a data group including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device.

[0064] A second acquisition step of acquiring operation content information that records the content of the maintenance operation performed by the maintenance worker on the target device, that is, the replaced or repaired parts or the new parts after replacement; and

[0065] A learning step of performing learning by associating the data group acquired in the first acquisition step with the replaced or repaired parts or the new parts after replacement recorded in the operation content information acquired in the second acquisition step.

[0066] According to the 15th aspect of the present disclosure, the parts to be replaced or repaired or the new parts after replacement in the target device can be determined before dispatch, and the preparations required for replacement or repair can be made before dispatch, so the number of dispatches of the maintenance workers can be reduced. In other words, it is possible to provide a maintenance operation assistance program that can assist the maintenance operation by reducing the number of dispatches of the maintenance workers.

[0067] In addition, the maintenance operation assistance program according to the 16th aspect of the present disclosure can cause a computer to execute the following steps, that is:

[0068] A first acquisition step of acquiring a data group including a combination of device information of the target device and operation information of the target device, or a data group including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device;

[0069] A second acquisition step of acquiring operation content information that records the content of the maintenance operation performed by the maintenance worker on the target device, that is, the operation process; and

[0070] A learning step of performing learning by associating the data group acquired in the first acquisition step with the operation process recorded in the operation content information acquired in the second acquisition step.

[0071] According to the 16th aspect of the present disclosure, the operation process of the maintenance operation for the target device can be determined before dispatch, and the preparations required for executing the operation process can be made before dispatch, so the number of dispatches of the maintenance workers can be reduced. In other words, according to the 16th aspect of the present disclosure, it is possible to provide a maintenance operation assistance program that can assist the maintenance operation by reducing the number of dispatches of the maintenance workers.

[0072] Furthermore, the maintenance operation assistance program according to the 17th aspect of the present disclosure is a maintenance operation assistance program for causing a computer to execute the following steps, that is:

[0073] A calculation step of calculating a reward based on evaluation information, which is information for evaluating the operation result of the maintenance operation for the target device; and

[0074] Learning step: learning about the components that should be replaced or repaired in the target device or the new components after replacement based on a data group that combines the device information of the target device and the operation information of the target device, or a data group that combines the device information of the target device and the phenomenon information representing the phenomenon related to the target device.

[0075] Among them, in the learning step, learning about the components that should be replaced or repaired in the target device or the new components after replacement according to the reward.

[0076] According to the 17th aspect of the present disclosure, before dispatch, it is possible to determine the components that should be replaced or repaired in the target device or the new components after replacement, and it is possible to make preparations required for replacement or repair before dispatch, so the number of dispatches of maintenance personnel can be reduced. In other words, according to the 17th aspect of the present disclosure, it is possible to provide a maintenance operation assistance program that can assist maintenance operations by reducing the number of dispatches of maintenance personnel.

[0077] In addition, the maintenance operation assistance program according to the 18th aspect of the present disclosure is a maintenance operation assistance program for causing a computer to execute the following steps, that is:

[0078] Calculation step: calculating a reward according to evaluation information, where the evaluation information is information for evaluating the operation result of the maintenance operation for the target device; and

[0079] Learning step: learning about the operation process of the maintenance operation for the target device based on a data group that combines the device information of the target device and the operation information of the target device, or a data group that combines the device information of the target device and the phenomenon information representing the phenomenon related to the target device.

[0080] Among them, in the learning step, learning about the operation process of the maintenance operation for the target device according to the reward.

[0081] According to the 18th aspect of the present disclosure, before dispatch, it is possible to determine the operation process of the maintenance operation for the target device, and it is possible to make preparations required for executing the operation process before dispatch, so the number of dispatches of maintenance personnel can be reduced. In other words, it is possible to provide a maintenance operation assistance program that can assist maintenance operations by reducing the number of dispatches of maintenance personnel. Brief Description of the Drawings

[0082] Figure 1 It is a schematic diagram of an example of the system configuration of the maintenance operation assistance system (learning stage).

[0083] Figure 2It is a schematic diagram of an example of device operation information and device phenomenon information.

[0084] Figure 3 It is a schematic diagram of an example of the functional configuration of the diagnostic unit of the monitoring device.

[0085] Figure 4 It is a schematic diagram of an example of device information, device user information, and maintenance process information.

[0086] Figure 5 It is a schematic diagram of a specific example of a maintenance process manual.

[0087] Figure 6 It is a schematic diagram of an example of job content information.

[0088] Figure 7 It is a schematic diagram of an example of the hardware structure of the maintenance work assistance device.

[0089] Figure 8 It is a schematic diagram of an example of the functions in the learning stage implemented in the maintenance work assistance device of the first embodiment.

[0090] Figure 9 It is a schematic diagram of an example of the system configuration of the maintenance work assistance system (inference stage).

[0091] Figure 10 It is the first diagram showing the details of the functional configuration of the inference unit.

[0092] Figure 11 It is a flowchart showing the process of the maintenance work assistance process performed by the maintenance work assistance device of the first embodiment.

[0093] Figure 12 It is a schematic diagram of an example of the functions in the learning stage implemented in the maintenance work assistance device of the second embodiment.

[0094] Figure 13 It is the second diagram showing the details of the functional configuration of the inference unit.

[0095] Figure 14 It is a flowchart showing the process of the maintenance work assistance process performed by the maintenance work assistance device of the second embodiment.

[0096] Figure 15 It is a schematic diagram of an example of the system configuration of the maintenance work assistance system (reinforcement learning stage).

[0097] Figure 16 It is a diagram showing the details of the functional configuration of the reinforcement learning unit.

[0098] Figure 17It is a flowchart showing the process of reinforcement learning processing performed on the maintenance operation assistance device of the third embodiment.

[0099] [Explanation of reference numerals]

[0100] 100: Maintenance operation assistance system; 110: Monitoring device; 120: Workstation; 130_1 to 130_n: Air conditioning equipment; 140: Maintenance device; 160: Maintenance operation assistance device; 161: Learning information acquisition unit; 162: Learning unit; 200: Equipment operation information; 210: Equipment phenomenon information; 400: Equipment information; 410: Equipment user information; 420: Maintenance process information; 600: Operation content information; 801: Component determination model; 802: Comparison and change unit; 910: Maintenance operation assistance device; 911: Equipment-related information acquisition unit; 912: Inference unit; 1000: Component determination model after learning; 1201: Operation content determination model; 1202: Comparison and change unit; 1310: Inference unit; 1320: Operation content determination model after learning; 1500: Maintenance operation assistance device; 1510: Reinforcement learning unit; 1520: Operation evaluation information acquisition unit; 1530: Reward calculation unit; 1600: Component determination model. Detailed implementation manners

[0101] The following describes each embodiment with reference to the drawings. It should be noted that components having substantially the same functional configuration are given the same reference numerals in this specification and the drawings, and thus redundant descriptions are omitted.

[0102] [First embodiment]

[0103] <System configuration of the maintenance operation assistance system (learning stage)>

[0104] First, the system configuration of the maintenance operation assistance system in the learning stage will be described. Figure 1 It is a schematic diagram of an example of the system configuration of the maintenance operation assistance system (learning stage). As Figure 1 shown, the maintenance operation assistance system 100 has, for example, a monitoring device 110, a workstation 120, air conditioning equipment 130_1 to 130_n, a maintenance device 140, and a maintenance operation assistance device 160. It should be noted that in the maintenance operation assistance system 100 in the learning stage, the monitoring device 110, the workstation 120, the maintenance device 140, and the maintenance operation assistance device 160 can be connected via the network 170.

[0105] The monitoring device 110 is a device that monitors the air conditioning equipment 130_1 to 130_n and sends a maintenance instruction to the maintenance device 140 when an abnormality occurs.

[0106] The monitoring device 110 is installed with a monitoring program. By executing this program, the monitoring device 110 can function as an operation and phenomenon information acquisition unit 111, a diagnosis unit 113, and a maintenance instruction sending unit 114.

[0107] The operation and phenomenon information acquisition unit 111 can obtain "equipment operation information" and / or "equipment phenomenon information" from air-conditioning equipment 130_1 to 130_n, etc. via the workstation 120 at a predetermined cycle, and save them in the operation and phenomenon information storage unit 112.

[0108] The equipment operation information refers to the information obtained from the air-conditioning equipment 130_1 to 130_n regardless of whether it is in the operation period or the stop period. The equipment operation information includes the abnormal code or abnormal prediction data output by the air-conditioning equipment when an abnormality occurs, the cumulative working time of the air-conditioning equipment, the operation information during the operation period (compressor speed, suction superheat degree, subcooling degree, etc.), the indoor temperature, the outdoor temperature, etc. It should be noted that the abnormal prediction data refers to the data output by a predetermined analysis software through analyzing the equipment operation information of the air-conditioning equipment 130_1 to 130_n and performing abnormal diagnosis.

[0109] In addition, the equipment phenomenon information refers to the information indicating the phenomena related to the air-conditioning equipment 130_1 to 130_n. The equipment phenomenon information not only includes the phenomena occurring in the air-conditioning equipment 130_1 to 130_n, but also includes the information indicating the influence on the indoor or outdoor due to the failure and / or abnormality of the air-conditioning equipment 130_1 to 130_n ("not cooling", "operation stop", "abnormal sound", etc.).

[0110] Regarding the diagnosis unit 113, when an abnormal code or abnormal prediction data is stored in the operation and phenomenon information storage unit 112, it can diagnose the cause of the abnormality based on other equipment operation information and / or equipment phenomenon information stored in the operation and phenomenon information storage unit 112. In addition, the diagnosis unit 113 can also notify the maintenance instruction sending unit 114 of the information indicating the diagnosed cause of the abnormality (abnormal cause information).

[0111] Alternatively, when an abnormal code or abnormal prediction data is stored in the operation and phenomenon information storage unit 112, the diagnosis unit 113 can predict the abnormal location by referring to a table (table) that has previously defined the correspondence between the abnormal code or abnormal prediction data and the abnormal location. In addition, the diagnosis unit 113 can also notify the maintenance instruction sending unit 114 of the information indicating the predicted abnormal location (abnormal location information).

[0112] Regarding the maintenance instruction sending unit 114, when notified of the abnormal cause information or abnormal location information from the diagnosis unit 113, this information can be sent to the maintenance device 140.

[0113] The workstation 120 is connected to the air conditioning devices 130_1 to 130_n, and can store the device operation information and / or device phenomenon information sent from the air conditioning devices 130_1 to 130_n at a predetermined cycle (for example, 10 seconds) in the internal memory. In addition, the workstation 120 can also send the device operation information and / or device phenomenon information stored in the internal memory to the monitoring device 110 at a predetermined cycle (for example, 1 hour).

[0114] The air conditioning devices 130_1 to 130_n are devices that work to remove air pollution in each room of the user's building and automatically adjust the temperature and humidity. The air conditioning devices 130_1 to 130_n can acquire device operation information during operation and send it to the workstation 120. In addition, the air conditioning devices 130_1 to 130_n can also acquire device phenomenon information during operation and send it to the workstation 120 as well.

[0115] A maintenance program is installed in the maintenance device 140. By executing this program, the maintenance device 140 can function as a maintenance instruction receiving unit 141, a maintenance information output unit 142, and a job content recording unit 146.

[0116] The maintenance instruction receiving unit 141 can receive the abnormal cause information or abnormal location information sent from the monitoring device 110. The maintenance instruction sending unit 114 can also notify the received abnormal cause information or abnormal location information to the maintenance information output unit 142.

[0117] Regarding the maintenance information output unit 142, when notified of the abnormal cause information or abnormal location information, it can read out the "device information" related to the target air conditioning device from the device information storage unit 143. The device information refers to information indicating the attributes of the air conditioning device. The device information can include device ID, device category (type), device capacity, power consumption (electricity consumption), installation (setting) years, etc.

[0118] In addition, when notified of the abnormal cause information or abnormal location information, the maintenance information output unit 142 can read out the "device user information" related to the user of the target air conditioning device from the device user information storage unit 144. The device user information refers to information related to the building used by the user of the air conditioning device. The device user information can include building use, building heat load, total floor area, building years, industry classification (type), etc.

[0119] In addition, regarding the maintenance information output unit 142, after being notified of the abnormality cause information or the abnormality location information, the "maintenance process information" that defines the processes of the maintenance work can also be read from the maintenance process information storage unit 145. The maintenance process information refers to the information that records the specific process content of the maintenance work (i.e., the specific content of each operation), and is the information classified and recorded according to each abnormality cause and / or abnormality location.

[0120] In addition, the maintenance information output unit 142 can notify the maintenance worker 150 of the read device information, device user information, and maintenance process information. Accordingly, the maintenance worker 150 can be aware that an abnormality has occurred in the air-conditioning equipment, and after grasping the abnormality cause, abnormality location, and the notified information, can go to the site where the air-conditioning equipment 130_1 to 130_n is installed.

[0121] It should be noted that the maintenance worker 150 conducts inspections based on the abnormality cause and / or abnormality location on the target air-conditioning equipment ( Figure 1 in the example is the air-conditioning equipment 130_n) at the site, thereby determining the faulty components to be replaced or repaired. Next, the maintenance worker 150 prepares tools for replacing or repairing the determined components and / or new components for replacement, and then goes to the site again. After that, the maintenance worker 150 stops the operating air-conditioning equipment 130_n and performs the work of repairing the faulty components or replacing them with new components. After the replacement or repair work of the components is completed, the maintenance worker 150 restarts the air-conditioning equipment 130_n and can record the content of the series of maintenance work in the maintenance work report 151. It should be noted that the maintenance work report 151 recorded by the maintenance worker 150 can be input into the maintenance device 140.

[0122] The operation content recording unit 146 can save the maintenance work content of the maintenance work report 151 input by the maintenance worker 150 in the operation content information storage unit 147.

[0123] The maintenance work assistance device 160 is a device that works in the learning stage. A maintenance work assistance program (learning stage) is installed in the maintenance work assistance device 160, and by executing this program, the maintenance work assistance device 160 can function as a learning information acquisition unit 161 and a learning unit 162.

[0124] The learning information acquisition unit 161 is an example of the first acquisition unit and the second acquisition unit, and can acquire learning information via the network 170. The learning information acquired by the learning information acquisition unit 161 includes the following information, etc., namely: · The device operation information and device phenomenon information saved in the operation and phenomenon information storage unit 112;

[0125] · The device information stored in the device information storage unit 143;

[0126] · The device user information stored in the device user information storage unit 144;

[0127] · The maintenance process information stored in the maintenance process information storage unit 145; and

[0128] · The job content information stored in the job content information storage unit 147.

[0129] Based on the learning information obtained by the learning information acquisition unit 161, the learning unit 162 can perform machine learning on the model for determining the parts to be replaced or repaired or the new parts after replacement. Accordingly, the learning unit 162 can generate a learned model (i.e., a well-learned model, also referred to as a trained model) for determining the parts to be replaced or repaired or the new parts after replacement.

[0130] <Explanation of device operation information and device phenomenon information>

[0131] Next, an explanation will be given of the device operation information and device phenomenon information stored in the operation and phenomenon information storage unit 112. Figure 2 It is a schematic diagram showing an example of the device operation information and device phenomenon information, where the information is stored separately for each air-conditioning device (for example, the device operation information 200 represents the device operation information and device phenomenon information of the air-conditioning device 130_1).

[0132] As Figure 2 shown in 2a, the items of information in the device operation information 200 include "time information", "abnormal code or abnormal prediction data", "accumulated working time", "operating conditions", "indoor temperature", and "outdoor temperature". In addition, the "operating conditions" also include "compressor speed", "suction superheat", "subcooling", "discharge temperature", "high and low pressure values", "high-pressure side temperature", "low-pressure side temperature", "valve opening", and "external heat exchange temperature".

[0133] The "time information" in the device operation information 200 stores the time when the air-conditioning device 130_1 obtains any one of "abnormal code or abnormal prediction data" to "outdoor temperature".

[0134] The content obtained by the air-conditioning device 130_1 is stored in "abnormal code or abnormal prediction data" to "outdoor temperature".

[0135] In addition, as Figure 2As shown in FIG. 2B, the items of information in the device phenomenon information 210 include "occurrence time" and "phenomenon". In addition, "phenomenon" includes not only the phenomena occurring in the air conditioning device, but also the impacts on the room due to the failure and / or abnormality of the air conditioning device ("no cooling", "operation stop", "higher electricity bill", etc.).

[0136] The "occurrence time" stores the occurrence time of any one of the phenomena included in the "phenomenon". Information indicating the occurrence of the phenomenon is stored separately for "no cooling", "operation stop", "higher electricity bill", etc.

[0137] <Details of the functional configuration of the diagnosis unit of the monitoring device>

[0138] Next, details of the functional configuration of the diagnosis unit 113 in each functional unit included in the monitoring device 110 will be described. Figure 3 It is a schematic diagram of an example of the functional configuration of the diagnosis unit of the monitoring device. As Figure 3 shown, the diagnosis unit 113 has a preprocessing unit 301, an abnormal cause diagnosis unit 302, and an abnormal position prediction unit 303.

[0139] The preprocessing unit 301 can read out the device operation information and the device phenomenon information from the operation and phenomenon information storage unit 112. Then, the preprocessing unit 301 can extract the information to be notified to the abnormal cause diagnosis unit 302 from the read device operation information and device phenomenon information, and obtain a part of the extracted information, or a value obtained according to the calculation result of calculating a part of the extracted information using a predetermined calculation formula.

[0140] For example, the preprocessing unit 301 can extract the operating conditions such as the outdoor temperature, the exhaust temperature, the high-pressure side temperature, and the valve opening from the device operation information 200. In addition, the preprocessing unit 301 can also obtain the time when these operating conditions or the calculation results of calculating these operating conditions using a predetermined calculation formula reach a predetermined threshold value. In addition, the preprocessing unit 301 can, for example, also extract the information indicating that the phenomenon = "no cooling" has occurred from the device phenomenon information. Furthermore, the preprocessing unit 301 can notify the extracted information or the obtained information to the abnormal cause diagnosis unit 302.

[0141] Alternatively, the preprocessing unit 301 can extract the information to be notified to the abnormal position prediction unit 303 from the read device operation information and device phenomenon information, and notify it to the abnormal position prediction unit 303.

[0142] For example, the preprocessing unit 301 can extract the abnormal code or the abnormal prediction data from the device operation information 200. In addition, the preprocessing unit 301 can also notify the extracted abnormal code or abnormal prediction data to the abnormal position prediction unit 303.

[0143] The abnormal cause diagnosis unit 302 can output the abnormal cause information by inputting the information notified from the preprocessing unit 301 into the abnormal cause table. In the abnormal cause table, the abnormal causes are defined as "gas shortage", "gas leakage", "heat exchange fouling", "air filter fouling", "exhaust pipe abnormality", "high pressure abnormality", etc. In addition, the diagnosis conditions for diagnosing each abnormal cause are also defined.

[0144] The abnormal cause diagnosis unit 302 can diagnose which abnormal cause it is by comparing the diagnosis conditions defined in the abnormal cause table with the information notified from the preprocessing unit 301, and output the abnormal cause information.

[0145] The abnormal location prediction unit 303 can output the abnormal location information by inputting the abnormal code or abnormal prediction data notified from the preprocessing unit 301 into the abnormal location table. In the abnormal location table, for each abnormal code or abnormal prediction code, the abnormal locations are defined as "electric valve", "compressor", "heat exchange", "refrigerant", "solenoid valve", etc. In addition, the confidence level is defined for each abnormal location.

[0146] The abnormal location prediction unit 303 can predict the abnormal location with a higher confidence level by referring to the abnormal location table corresponding to the abnormal code or abnormal prediction data, and output it as the abnormal location information.

[0147] <Explanation of Equipment Information, Equipment User Information, and Maintenance Process Information>

[0148] Next, the equipment information stored in the equipment information storage unit 143, the equipment user information stored in the equipment user information storage unit 144, and the maintenance process information stored in the maintenance process information storage unit 145 will be explained.

[0149] Figure 4 is a schematic diagram of an example of equipment information, equipment user information, and maintenance process information. The equipment information is stored separately for each air conditioning equipment, the equipment user information is stored separately for each building where the air conditioning equipment is installed, and the maintenance process information is stored separately for each type (category) of air conditioning equipment (for example, the equipment information 400 represents the equipment information of the air conditioning equipment 130_1).

[0150] As Figure 4 shown in 4a, the items of information in the equipment information 400 include "equipment ID", "equipment category (type)", "equipment capacity", "power consumption", and "installation (setup) years (elapsed years)". The equipment information 400 can be stored in the equipment information storage unit 143 when the air conditioning equipment 130_1 is installed.

[0151] In addition, as shown in Figure 4 4b, the items of information in the equipment user information 410 include "building use", "building heat load", "total floor area", "age of building", and "type of business". The equipment user information 410 can be stored in the equipment user information storage unit 144, for example, when the air conditioning equipment 130_1 to 130_n is installed.

[0152] In addition, as shown in Figure 4 4c, the items of information in the maintenance process information 420 include "cause of abnormality", "location of abnormality", and "maintenance process guide". The maintenance process information 420 is information corresponding to the equipment category (type) of the air conditioning equipment 130_1, and the corresponding maintenance process guides are stored according to each cause of abnormality and location of abnormality.

[0153] It should be noted that Figure 4 in the example of 4c, it shows a case where "abnormal high pressure" is stored as the "cause of abnormality", "outdoor unit" is stored as the "location of abnormality", and "maintenance process guide 1" is stored as the "maintenance process guide". In addition, Figure 4 in the example of 4c, for the "location of abnormality" = "outdoor unit", only one maintenance process guide is shown, but actually multiple maintenance process guides can be stored.

[0154] <Specific Example of Maintenance Process Guide>

[0155] Next, a specific example of the maintenance process guide stored in the maintenance process information storage unit 145 will be described. Figure 5 is a diagram showing a specific example of the maintenance process guide, which shows "maintenance process guide 1". As described above, the maintenance process guide 1 is used for the case where the cause of abnormality is abnormal high pressure and the location of abnormality is the outdoor unit.

[0156] The maintenance worker 150 performs operations according to the operation process 500. In the case of the operation process 500, the high pressure switch, high pressure sensor, and substrate are inspected in sequence. For example, when it is determined that the high pressure switch is abnormal, the high pressure switch is determined as the component to be replaced (see operation step 501). In addition, when it is determined that the high pressure sensor is abnormal, the high pressure sensor is determined as the component to be replaced (see operation step 502). In addition, when it is determined that the substrate is abnormal, the substrate is determined as the component to be replaced (see operation step 503).

[0157] Note that, when no abnormality occurs in any component, the maintenance operator 150 proceeds to the next maintenance process guide ("Maintenance Process Guide 2") and performs operations according to the operation process described in the next maintenance process guide.

[0158] <Explanation of operation content information>

[0159] Next, the operation content information stored in the operation content information storage unit 147 will be explained. Figure 6 It is a schematic diagram of an example of operation content information. As Figure 6 shown, the items of information in the operation content information 600 include "operation date and time", "operation time used", "operation personnel", "target equipment ID", "abnormality cause", "abnormality location", "maintenance process guide", "component before replacement", "component after replacement", and "site situation".

[0160] The "operation date and time" stores the date and start time when the maintenance operator 150 performs maintenance operations on-site. The "operation time used" stores the time required for the maintenance operator 150 to perform maintenance operations on-site. The "operation personnel" stores the identifier used to identify (label) the operation personnel who perform maintenance operations on-site.

[0161] The "target equipment ID" stores the identifier used to identify the air-conditioning equipment that is the processing target for the maintenance operator 150 to perform maintenance operations on-site.

[0162] The "abnormality cause" stores the component information related to the cause of the abnormality determined by the maintenance operator 150 through actual inspection of the air-conditioning equipment on-site. The "abnormality location" stores the component information related to the location of the abnormality determined by the maintenance operator 150 through actual inspection of the air-conditioning equipment on-site.

[0163] The "component before replacement" stores the component information (component code) related to the faulty component that has been replaced or repaired as determined by the maintenance operator 150 through actual inspection of the air-conditioning equipment on-site. The "component after replacement" stores the information (component code) related to the new component after replacement that has been replaced by the maintenance operator 150.

[0164] The "maintenance process guide" stores the information used to identify the maintenance process guide used by the maintenance operator 150 when determining the faulty component (for example, Maintenance Process Guide 1, etc.).

[0165] The "site situation" stores the information related to the site situation noticed by the maintenance operator 150. The information related to the site condition refers to, for example,

[0166] · Abnormal sound appears in the air-conditioning equipment

[0167] · Objects are placed near the air conditioning equipment.

[0168] · The actual outside air temperature at the site is higher than the outside air temperature detected by the air conditioning equipment.

[0169] Information that cannot be obtained from the equipment operation information and / or equipment phenomenon information, such as the above.

[0170] <Hardware Structure of Maintenance Work Support Device (Learning Stage)>

[0171] Next, the hardware structures of the respective devices constituting the maintenance work support system 100 will be described. It should be noted that here, only the hardware structure of the maintenance work support device 160 in the learning stage will be described as a representative.

[0172] Figure 7 is a schematic diagram of an example of the hardware structure of the maintenance work support device. As Figure 7 shown, the maintenance work support device 160 includes a CPU (Central Processing Unit) 701, a ROM (Read Only Memory) 702, and a RAM (Random Access Memory) 703. The CPU 701, ROM 702, and RAM 703 can form a so-called computer. In addition, the maintenance work support device 160 also includes an auxiliary storage device 704, a display device 705, an operation device 706, an I / F (Interface) device 707, and a drive device 708. Each hardware of the maintenance work support device 160 can be interconnected via a bus 709.

[0173] The CPU 701 is a computing device that can execute various programs (for example, maintenance work support program (learning stage), etc.) installed in the auxiliary storage device 704. The ROM 702 is a non-volatile memory. The ROM 702 functions as a main storage device and can save various programs, data, etc. required for the CPU 701 to execute various programs installed in the auxiliary storage device 704. Specifically, the ROM 702 can save boot programs such as BIOS (Basic Input / Output System) and EFI (Extensible Firmware Interface).

[0174] The RAM 703 is a volatile memory such as DRAM (Dynamic Random Access Memory) or SRAM (Static Random Access Memory). The RAM 703 functions as a main storage device and can provide a working space for various programs installed in the auxiliary storage device 704 to be executed by the CPU 701.

[0175] The auxiliary storage device 704 can store various programs, information used when executing various programs, and so on.

[0176] The display device 705 is a display device that displays the internal state of the maintenance operation assistance device 160. The operation device 706 is, for example, an operation device for enabling the administrator of the maintenance operation assistance device 160 to perform various operations on the maintenance operation assistance device 160. The I / F device 707 is a communication device for communicating by connecting to the network 170.

[0177] The drive device 708 is a device for inserting the recording medium 710. The recording medium 710 mentioned here includes CD-ROMs, floppy disks, and media that record information in optical, electrical, or magnetic ways such as optical disks. In addition, the recording medium 710 also includes ROMs, semiconductor memories that record information electrically such as flash memories, and so on.

[0178] It should be noted that, for example, for various programs installed in the auxiliary storage device 704, they can be installed by inserting the distributed recording medium 710 into the drive device 708 and having the drive device 708 read out the various programs recorded in the recording medium 710. Or, various programs installed in the auxiliary storage device 704 can also be installed by downloading from the network 170.

[0179] <Functional Composition of Maintenance Operation Assistance Device (Learning Phase)>

[0180] Next, the functional composition of the maintenance operation assistance device 160 in the learning phase will be described. Figure 8 It is a schematic diagram of an example of the functions in the learning phase implemented in the maintenance operation assistance device of the first embodiment. As Figure 8 shown, the maintenance operation assistance device 160 has a learning information acquisition unit 161 and a learning unit 162.

[0181] The learning information acquisition unit 161 can acquire the information (learning information) used by the learning unit 162 for machine learning. Specifically, the learning information acquisition unit 161 can acquire device operation information and device phenomenon information from the operation and phenomenon information storage unit 112 of the monitoring device 110. In addition, the learning information acquisition unit 161 can also acquire device information and device user information from the device information storage unit 143 and the device user information storage unit 144 of the maintenance device 140. Additionally, the learning information acquisition unit 161 can acquire operation content information from the operation content information storage unit 147 of the maintenance device 140. Furthermore, the learning information acquisition unit 161 can notify the acquired learning information to the learning unit 162.

[0182] The learning unit 162 includes a component determination model 801 and a comparison and change unit 802. By inputting the device operation information, device phenomenon information, device information, and device user information in the notified learning information into the component determination model 801, the learning unit 162 can cause the component determination model 801 to execute. Accordingly, the component determination model 801 can output component information (component code).

[0183] The component information (component code) output from the component determination model 801 can be input into the comparison and change unit 802. The comparison and change unit 802 can

[0184] · the component information (component code) output from the component determination model 801

[0185] and

[0186] · the component information (component code) (ground truth data) notified from the learning information acquisition unit 161

[0187] be compared. Accordingly, the comparison and change unit 802 can change the model parameters of the component determination model 801 based on the comparison result. It should be noted that the component information notified from the learning information acquisition unit 161 refers to a failed component (pre-replacement component) that has been replaced or repaired, or a new component (post-replacement component) after replacement.

[0188] In this way, the learning unit 162 can

[0189] · the device operation information, device phenomenon information, device information, and device user information (data set)

[0190] and

[0191] · the component determination model 801 for determining the correspondence between the component information (component code) perform machine learning. Accordingly, the learning unit 162 can generate a learned component determination model for determining the component information (component code).

[0192] It should be noted that Figure 8 In the example of , although it shows the case where the learning unit 162 inputs the device operation information, device phenomenon information, device information, and device user information into the component determination model 801, only a part of this information can also be input into the component determination model 801.

[0193] For example, the learning unit 162 can be used to determine

[0194] · Device operation information and device information (data group)

[0195] and

[0196] · Component information (component code) corresponding relationship component determination model 801, or, used to determine

[0197] · Device phenomenon information and device information (data group)

[0198] and

[0199] · Component information (component code) corresponding relationship component determination model 801 for machine learning. Accordingly, the learning unit 162 can also generate a learned component determination model for determining component information (component code).

[0200] <System configuration of the maintenance operation assistance system (inference stage)>

[0201] Next, the system configuration of the maintenance operation assistance system in the inference stage will be described. Figure 9 is a schematic diagram of an example of the system configuration of the maintenance operation assistance system (inference stage). The difference from the maintenance operation assistance system 100 in the learning stage ( Figure 1 ) is that in the case of the maintenance operation assistance system 900 in the inference stage, it does not have the maintenance operation assistance device 160 in the learning stage; in addition, in the case of the maintenance operation assistance system 900 in the inference stage, in addition to the monitoring device 110, it also has the maintenance operation assistance device 910 in the inference stage.

[0202] The maintenance operation assistance device 910 is a device that monitors the air-conditioning equipment 130_1 to 130_n and sends a maintenance instruction to the maintenance device 140 when an abnormality occurs.

[0203] A maintenance operation assistance program (inference stage) is installed in the maintenance operation assistance device 910. By executing this program, the maintenance operation assistance device 910 can function as an operation and phenomenon information acquisition unit 111, a device-related information acquisition unit 911, an inference unit 912, and a maintenance instruction sending unit 114.

[0204] It should be noted that since it has been referred to through Figure 1The operation and phenomenon information acquisition unit 111 and the maintenance instruction sending unit 114 have been described, so their descriptions are omitted here.

[0205] The equipment-related information acquisition unit 911 can acquire the information (equipment-related information) used by the inference unit 912 for inference processing from the maintenance device 140. Specifically, the equipment-related information acquisition unit 911 can acquire equipment information and equipment user information from the equipment information storage unit 143 and the equipment user information storage unit 144 of the maintenance device 140. In addition, the equipment-related information acquisition unit 911 can also notify the acquired equipment-related information (equipment information and equipment user information) to the inference unit 912.

[0206] The inference unit 912 has a learned component determination model. By executing the learned component determination model, component information (component code) can be output. The inference unit 912 performs

[0207] · the equipment operation information and equipment phenomenon information stored in the operation and phenomenon information storage unit 112

[0208] and

[0209] · the input of the equipment-related information (equipment information and equipment user information) notified from the equipment-related information acquisition unit 911, which can execute the learned component model. It should be noted that in the inference stage, the maintenance instruction sending unit 114 can send the component information (component code) to the maintenance device 140.

[0210] <Details of the functional composition of the inference unit>

[0211] Next, the details of the functional composition of the inference unit 912 in the functional composition of the maintenance operation assistance device 910 in the inference stage will be described. Figure 10 This is the first figure showing the details of the functional composition of the inference unit. As Figure 10 shown, the inference unit 912 has a learned component determination model 1000.

[0212] The inference unit 912 can read out the equipment operation information and equipment phenomenon information stored in the operation and phenomenon information storage unit 112. It should be noted that the equipment operation information and equipment phenomenon information read out by the inference unit 912 are different from the equipment operation information and equipment phenomenon information read out when the learning unit 162 performs machine learning (that is, the equipment operation information and equipment phenomenon information are read again).

[0213] In addition, the inference unit 912 can also acquire the equipment information and equipment user information notified from the equipment-related information acquisition unit 911.

[0214] The inference unit 912 can cause the learned component determination model 1000 to execute by inputting the read device operation information, device phenomenon information, obtained device information, and device user information into the learned component determination model 1000. Accordingly, the learned component determination model 1000 can infer component information (component code).

[0215] In this way, by inferring the component information (component code) based on the current device operation information, device phenomenon information, device information of the air-conditioning device as the processing target, and device user information, the inference unit 912 can infer appropriate component information (component code).

[0216] It should be noted that in the above description, the inference unit 912 inputs the device operation information, device phenomenon information, device information, and device user information into the learned component determination model 1000. However, when the learned component determination model 1000 is generated based on the device operation information and device information, the inference unit 912 can input only the device operation information and device information into the learned component determination model 1000. In addition, when the learned component determination model 1000 is generated based on the device phenomenon information and device information, the inference unit 912 can input only the device phenomenon information and device information into the learned component determination model 1000.

[0217] In addition, the component information (component code) inferred by the inference unit 912 is the component information (component code) of the component to be replaced or repaired, or the new component after replacement.

[0218] <Process flow of maintenance operation assistance processing>

[0219] Next, the process flow of the maintenance operation assistance processing performed by the maintenance operation assistance device 160 in the learning stage and the maintenance operation assistance device 910 in the inference stage will be described. Figure 11 It is a flowchart showing the process flow of the maintenance operation assistance processing performed by the maintenance operation assistance device of the first embodiment.

[0220] In step S1101, the learning information acquisition unit 161 acquires device operation information and device phenomenon information.

[0221] In step S1102, the learning information acquisition unit 161 acquires operation content information.

[0222] In step S1103, the learning unit 162 causes the component determination model 801 to execute by inputting the device operation information and the device phenomenon information into the component determination model 801. Next, the learning unit 162 performs machine learning on the component determination model 801 in such a way that the component information (component code) output by the component determination model 801 approaches the component information (component code) (correct data) included in the operation content information acquired by the learning information acquisition unit 161. Accordingly, the learning unit 162 can generate a learned component determination model. It should be noted that the generated learned component determination model can be embedded in the maintenance work support device 910 in the inference stage.

[0223] In step S1104, the inference unit 912 acquires the device operation information and the device phenomenon information of the air conditioner device that is the target device.

[0224] In step S1105, the inference unit 912 causes the learned component determination model 801 to execute by inputting the acquired device operation information and device phenomenon information into the learned component determination model 801. Accordingly, the inference unit 912 can infer the component information (component code).

[0225] In step S1106, the inference unit 912 determines whether to end the maintenance work support process. When it is determined in step S1106 that the maintenance work support process is to continue (in the case of NO (No) in step S1106), the process returns to step S1104.

[0226] On the other hand, when it is determined in step S1106 that the maintenance work support process is to end (in the case of YES (Yes) in step S1106), the maintenance work support process ends.

[0227] It should be noted that Figure 11 shows a case where the learning unit 162 performs collaborative learning to change the model parameters by inputting the device operation information and the device phenomenon information into the component determination model 801 together. However, the learning unit 162 can also perform sequential learning to change the model parameters by inputting the device operation information and the device phenomenon information into the component determination model 801 in a predetermined quantity in sequence.

[0228] <Summary>

[0229] As can be seen from the above description, the maintenance work support device of the first embodiment can

[0230] · Learn by associating the device operation information and the device information (or the device phenomenon information and the device information) with the component information (component code), and can

[0231] ·Based on the learning results, component information (component code) is inferred from the newly obtained device operation information and device information (or device phenomenon information and device information).

[0232] Accordingly, with the maintenance operation assistance device according to the first embodiment, when an abnormality occurs, it is possible to infer component information (component code) indicating a faulty component to be replaced or repaired or a new component after replacement.

[0233] In this way, according to the first embodiment, before the maintenance personnel are dispatched, the component information (component code) can be determined, and the preparations required for replacement or repair can be made before the dispatch, so the number of dispatches can be reduced. In other words, according to the first embodiment, it is possible to provide a maintenance operation assistance device, a maintenance operation assistance method, and a maintenance operation assistance program that can assist in maintenance operations by reducing the number of dispatches of maintenance personnel.

[0234] [Second Embodiment]

[0235] In the above first embodiment, the case where machine learning is performed using the component information (component code) in the operation content information as correct answer data has been described. In the second embodiment, the case where machine learning is performed using the maintenance process guide information in the operation content information as correct answer data will be described. Below, the second embodiment will be described centering on the differences from the above first embodiment.

[0236] <Functional Configuration of Maintenance Operation Assistance Device (Learning Phase)>

[0237] First, the functional configuration of the maintenance operation assistance device 160 in the learning phase of the second embodiment will be described. Figure 12 It is a schematic diagram of an example of the functions in the learning phase implemented in the maintenance operation assistance device of the second embodiment. Different from Figure 8 The difference is that in Figure 12 In this case, the learning unit 162 has an operation content determination model 1201 and a comparison and change unit 1202. By inputting the device operation information, device phenomenon information, device information, and device user information in the learning information notified to the operation content determination model 1201, the learning unit 162 can cause the operation content determination model 1201 to execute. Accordingly, the operation content determination model 1201 can output maintenance process guide information.

[0238] The maintenance process guide information output by the operation content determination model 1201 can be input to the comparison and change unit 1202. Accordingly, the comparison and change unit 1202 can

[0239] ·The maintenance process guide information output by the operation content determination model 1201

[0240] and

[0241] · The maintenance process guide information (correct data) notified from the learning information acquisition unit 161

[0242] is compared. In this way, the comparison and change unit 1202 can change the model parameters of the operation content determination model 1201 according to the comparison result.

[0243] It should be noted that the maintenance process guide information notified from the learning information acquisition unit 161 refers to, for example, information for determining the maintenance process guide such as "Maintenance Process Guide 1". Or, it may also be the operation process included in the maintenance process guide (for example, operation process 500).

[0244] In this way, the learning unit 162 can determine the

[0245] · Equipment operation information, equipment phenomenon information, equipment information, and equipment user information (data set)

[0246] and

[0247] · The operation content determination model 1201 for the correspondence relationship with the maintenance process guide information performs machine learning. Accordingly, the learning unit 162 can generate a learned operation content determination model for determining the maintenance process guide information.

[0248] It should be noted that Figure 12 Although the example of shows the case where the learning unit 162 inputs the equipment operation information, equipment phenomenon information, equipment information, and equipment user information into the operation content determination model 1201, only a part of this information may be input into the operation content determination model 1201.

[0249] For example, the learning unit 162 can determine the

[0250] · Equipment operation information and equipment information (data set)

[0251] and

[0252] · The operation content determination model 1201 for the correspondence relationship with the maintenance process guide information, or, for determining the

[0253] · Equipment phenomenon information and equipment information (data set)

[0254] and

[0255] · The operation content determination model 1201 for the correspondence relationship with the maintenance process guide information performs machine learning. Accordingly, the learning unit 162 can generate a learned operation content determination model for determining the maintenance process guide information.

[0256] <Details of the functional composition of the inference unit>

[0257] Next, the details of the functional composition of the inference unit in the functional composition of the maintenance operation assistance device 910 in the inference stage will be described. Figure 13 This is the second figure showing the details of the functional composition of the inference unit. Different from Figure 10 In the case where Figure 13 is different, the inference unit 1310 has a learned operation content determination model 1320.

[0258] By inputting the read device operation information, device phenomenon information, and the obtained device information and device user information into the learned operation content determination model 1320, the inference unit 1310 can cause the learned operation content determination model 1320 to execute. Accordingly, the learned operation content determination model 1320 can perform the inference of the maintenance process guide information.

[0259] In this way, by inferring the maintenance process guide information based on the current device operation information, device phenomenon information, and the device information and device user information of the air-conditioning device to be processed, the inference unit 1310 can infer appropriate maintenance process guide information.

[0260] It should be noted that in the above description, the inference unit 1310 inputs the device operation information, device phenomenon information, and device information and device user information into the learned operation content determination model 1320. However, when the learned operation content determination model 1320 is generated based on the device operation information and device information, the inference unit 1310 can input only the device operation information and device information into the learned operation content determination model 1320. In addition, when the learned operation content determination model 1320 is generated based on the device phenomenon information and device information, the inference unit 1310 can input only the device phenomenon information and device information into the learned operation content determination model 1320.

[0261] In addition, the maintenance process guide information inferred by the inference unit 1310 is information for determining the maintenance process guide or the operation process included in the maintenance process guide.

[0262] <Process of the maintenance operation assistance process>

[0263] Next, the process of the maintenance operation assistance process performed by the maintenance operation assistance device 160 in the learning stage and the maintenance operation assistance device 910 in the inference stage will be described. Figure 14 This is a flowchart showing the process of the maintenance operation assistance process performed by the maintenance operation assistance device of the second embodiment. Different from the flowchart shown in Figure 11 the differences are in steps S1401 and S1402.

[0264] In step S1401, the learning unit 162 causes the operation content determination model 1201 to execute by inputting the device operation information and the device phenomenon information into the operation content determination model 1201. Then, the learning unit 162 performs machine learning on the operation content determination model 1201 in such a way that the maintenance process guide information output by the operation content determination model 1201 approaches the maintenance process guide information (correct data) included in the acquired maintenance process information. Accordingly, the learning unit 162 can generate a learned operation content determination model. It should be noted that the generated learned operation content determination model can be embedded in the maintenance work support device 910 in the inference stage.

[0265] In step S1402, the inference unit 912 causes the learned operation content determination model 1320 to execute by inputting the acquired device operation information and device phenomenon information into the learned operation content determination model 1320. Accordingly, the inference unit 912 can infer the maintenance process guide information indicating the maintenance process that the maintenance worker 150 should execute.

[0266] <Summary>

[0267] As can be seen from the above description, the maintenance work support device according to the second embodiment can

[0268] · Learn by associating device operation information and device information (or device phenomenon information and device information) with operation process information,

[0269] and can

[0270] · According to the learning result, infer the maintenance process guide information from the newly acquired device operation information and device information (or device phenomenon information and device information).

[0271] Accordingly, with the maintenance work support device according to the second embodiment, when an abnormality occurs, it is possible to infer the maintenance process guide information indicating the maintenance process that the maintenance worker should execute.

[0272] In this way, according to the second embodiment, the maintenance process guide information can be determined before departure, and the preparations required for executing the operation process can be made before departure, so the number of departures of the maintenance workers can be reduced. In other words, according to the second embodiment, it is possible to provide a maintenance work support device, a maintenance work support method, and a maintenance work support program that can assist the maintenance work by reducing the number of departures of the maintenance workers.

[0273] [Third Embodiment]

[0274] In the above first embodiment, the case of performing machine learning on the component determination model using correct answer data for machine learning was described. In the third embodiment, the case of performing reinforcement learning on the component determination model will be described. Below, the third embodiment will be described centering on the differences from the above first embodiment.

[0275] <System configuration of maintenance work support system (reinforcement learning stage)>

[0276] First, the system configuration of the maintenance work support system will be described. Figure 15 It is a schematic diagram of an example of the system configuration of the maintenance work support system (reinforcement learning stage). Different from Figure 1 The difference lies in the maintenance work support device 1500.

[0277] A maintenance work support program is installed in the maintenance work support device 1500. By executing this program, the maintenance work support device 1500 can function as an operation and phenomenon information acquisition unit 111, a reinforcement learning unit 1510, a maintenance instruction sending unit 114, a work evaluation information acquisition unit 1520, and a reward calculation unit 1530.

[0278] Regarding the operation and phenomenon information acquisition unit 111 and the maintenance instruction sending unit 114, since they have been described in the above first embodiment by using Figure 1 they are not described here.

[0279] The reinforcement learning unit 1510 can read out the device operation information and device phenomenon information from the operation and phenomenon information storage unit 112, and perform reinforcement learning on the model for determining the components to be replaced or repaired or the new components after replacement. The reinforcement learning unit 1510 performs reinforcement learning in such a way that the reward output from the reward calculation unit 1530 becomes maximum. In addition, the reinforcement learning unit 1510 can also send the component information (component code) obtained through reinforcement learning to the maintenance instruction sending unit 114.

[0280] The work evaluation information acquisition unit 1520 can obtain the work evaluation information for calculating the reward from the maintenance device 140 via the network 170. The work evaluation information refers to the information for evaluating the work result of the maintenance work on the air-conditioning equipment as the processing target. In addition, the work evaluation information acquisition unit 1520 can also notify the obtained work evaluation information to the reward calculation unit 1530.

[0281] The reward calculation unit 1530 is an example of a calculation unit, and can calculate the reward used when the reinforcement learning unit 1510 performs reinforcement learning based on the work evaluation information.

[0282] <Details of the functional configuration of the reinforcement learning unit>

[0283] Next, the details of the functional configuration of the reinforcement learning unit 1510 will be described. Figure 16 This is a diagram showing the details of the functional configuration of the reinforcement learning unit.

[0284] As Figure 16 shown, the reinforcement learning unit 1510 has a component determination model 1600. The reinforcement learning unit 1510 changes the model parameters of the component determination model 1600 in such a way as to maximize the reward calculated by the reward calculation unit 1530. In addition, the reinforcement learning unit 1510 can execute the component determination model 1600 by inputting the device operation information, device phenomenon information, device information, and device user information read from the operation and phenomenon information storage unit 112 into the component determination model 1600 with the changed model parameters. Accordingly, the component determination model 1600 can output component information.

[0285] In this way, the reinforcement learning unit 1510 can perform reinforcement learning on the component determination model 1600 in such a way as to maximize the reward calculated based on the operation evaluation information when performing maintenance operations according to the previous component information. Accordingly, the reinforcement learning unit 1510 can output appropriate component information.

[0286] <Process of Reinforcement Learning>

[0287] Next, the process of the reinforcement learning performed by the maintenance operation support device 1500 of the third embodiment will be described. Figure 17 This is a flowchart showing the process of the reinforcement learning performed by the maintenance operation support device of the third embodiment.

[0288] In step S1701, the reinforcement learning unit 1510 acquires device operation information and device phenomenon information.

[0289] In step S1702, the operation evaluation information acquisition unit 1520 acquires operation evaluation information.

[0290] In step S1703, the reward calculation unit 1530 calculates a reward based on the operation evaluation information.

[0291] In step S1704, the reward calculation unit 1530 determines whether the calculated reward is greater than or equal to a predetermined threshold. If it is determined in step S1704 that the calculated reward is less than the predetermined threshold (the case of "no" in step S1704), the process proceeds to step S1705.

[0292] In step S1705, the reinforcement learning unit 1510 performs machine learning on the component determination model 1600 in such a way as to maximize the calculated reward.

[0293] In step S1706, the reinforcement learning unit 1510 can cause the component determination model 1600 to execute by inputting the acquired device operation information and device phenomenon information into the component determination model 1600. Accordingly, the reinforcement learning unit 1510 can output component information.

[0294] In step S1701, the maintenance instruction sending unit 114 sends the component information to the maintenance device 140 and returns to step S1701.

[0295] On the other hand, in the case where it is determined in step S1704 that the calculated reward is greater than or equal to a predetermined threshold (the case where it is "yes" in step S1704), the reinforcement learning process ends.

[0296] <Summary>

[0297] As can be seen from the above description, the maintenance work support device of the third embodiment can

[0298] · Calculate the reward for the component information based on the operation evaluation information. In addition, it is also possible to perform reinforcement learning on the component determination model in such a way that the calculated reward is maximized.

[0299] And it can

[0300] · By inputting the device operation information and device phenomenon information, cause the component determination model that has undergone reinforcement learning to execute, and thereby output component information (component code).

[0301] Accordingly, according to the maintenance work support device of the third embodiment, when an abnormality occurs, it is possible to output component information (component code) indicating the component that should be replaced or repaired or the new component after replacement.

[0302] In this way, according to the third embodiment, before the maintenance staff is dispatched, the component information (component code) can be determined, and the preparations required for replacement or repair can be made before dispatch, so the number of dispatches can be reduced. In other words, according to the third embodiment, it is possible to provide a maintenance work support device, a maintenance work support method, and a maintenance work support program that can assist in maintenance work by reducing the number of dispatches of maintenance staff.

[0303] [Other Embodiments]

[0304] In the above-described second embodiment, the case where positive solution data is used for machine learning when performing machine learning on the operation content determination model has been described. However, the method of machine learning is not limited to this, and even for the operation content determination model, it can be configured to perform machine learning by reinforcement learning in the same manner as the above-described third embodiment.

[0305] In each of the above-described embodiments, although the details of the models (component determination model and job content determination model) used in machine learning are not specifically mentioned, any type of model can be adopted for the models used in machine learning. Specifically, any type of model such as an NN (Neural Network) model, a random forest model, an SVM (Support Vector Machine) model, etc. can be used.

[0306] In addition, in the above-described first and second embodiments, although the details of the change method in the case of changing the model parameters based on the comparison result of the comparison and change unit are not specifically mentioned, regarding the change method of the model parameters performed by the comparison and change unit, it can be appropriately selected according to the type of the model.

[0307] Furthermore, in the above-described third embodiment, although the details of the calculation method of the reward performed by the reward calculation unit are not specifically mentioned, any method can be adopted for the calculation method of the reward performed by the reward calculation unit.

[0308] Although the embodiments have been described above, various changes and modifications can be made thereto as long as they do not depart from the gist and scope of the claims.

[0309] This application claims the priority based on Japanese Patent Application No. 2019-052019 filed on March 19, 2019, and incorporates the entire content of the Japanese patent application by reference into this application.

Claims

1. A maintenance operation assistance device, comprising: A first acquisition unit that acquires a data set including a combination of device information of a target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device; A second acquisition unit that acquires operation content information, in which the content of the maintenance operation performed by the maintenance operator on the target device, that is, the replaced or repaired component or the new component after replacement, is recorded; and A learning unit that learns in such a way as to associate the data set acquired by the first acquisition unit with the replaced or repaired component or the new component after replacement recorded in the operation content information acquired by the second acquisition unit.

2. A maintenance operation assistance device, comprising: A first acquisition unit that acquires a data set including a combination of device information of a target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device; A second acquisition unit that acquires operation content information, in which the content of the maintenance operation performed by the maintenance operator on the target device, that is, the operation process, is recorded; and A learning unit that learns in such a way as to associate the data set acquired by the first acquisition unit with the operation process recorded in the operation content information acquired by the second acquisition unit.

3. A maintenance operation assistance device, comprising: A calculation unit that calculates a reward based on evaluation information, which is information for evaluating the operation result of the maintenance operation on the target device; and A learning unit that learns about the component to be replaced or repaired or the new component after replacement in the target device based on a data set including a combination of device information of the target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device, Among them, The learning unit learns about the component to be replaced or repaired or the new component after replacement in the target device based on the reward.

4. A maintenance operation assistance device, comprising: A calculation unit that calculates a reward based on evaluation information, which is information for evaluating the operation result of the maintenance operation on the target device; and A learning unit that learns about the operation process of the maintenance operation on the target device based on a data set including a combination of device information of the target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device, Among them, The learning unit learns about the operation process of the maintenance operation on the target device based on the reward.

5. The maintenance operation assistance device according to any one of claims 1 to 4, wherein The device information includes the category of the target device.

6. The maintenance operation assistance device according to any one of claims 1 to 4, wherein The operation information includes any one of an error code output by the target device, anomaly prediction data, operating conditions, indoor temperature, outdoor temperature, and cumulative working hours.

7. The maintenance operation assistance device according to claim 6, wherein the target device is an air conditioning device, the operating conditions include any one of the compressor speed, the suction superheat degree, the subcooling degree, the exhaust temperature, the high and low pressure values, the high-pressure side temperature, the low-pressure side temperature, the valve opening degree, and the external heat exchange temperature.

8. The maintenance operation assistance device according to any one of claims 1 to 4, wherein the phenomenon information includes information indicating a phenomenon occurring due to a failure and / or abnormality of the target device.

9. The maintenance operation assistance device according to claim 1, further comprising: an inference unit that re-acquires a data set including a combination of device information of the target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device, and infers a component to be replaced or repaired in the target device or a new component after replacement from the re-acquired data set according to the result of learning by the learning unit.

10. The maintenance operation assistance device according to claim 2, further comprising: an inference unit that re-acquires a data set including a combination of device information of the target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device, and infers an operation process of a maintenance operation for the target device from the re-acquired data set according to the result of learning by the learning unit.

11. A maintenance operation assistance method, comprising: a first acquisition step of acquiring a data set including a combination of device information of the target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device; a second acquisition step of acquiring operation content information that records the content of the maintenance operation performed by the maintenance operator on the target device, namely, the component that has been replaced or repaired or the new component after replacement; and a learning step of learning in such a manner that the data set acquired in the first acquisition step is associated with the component that has been replaced or repaired or the new component after replacement recorded in the operation content information acquired in the second acquisition step.

12. A maintenance operation assistance method, comprising: a first acquisition step of acquiring a data set including a combination of device information of the target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device; a second acquisition step of acquiring operation content information that records the content of the maintenance operation performed by the maintenance operator on the target device, namely, the operation process; and a learning step of learning in such a manner that the data set acquired in the first acquisition step is associated with the operation process recorded in the operation content information acquired in the second acquisition step.

13. A maintenance operation assistance method, comprising: A calculation step of calculating a reward based on evaluation information which is information for evaluating the result of a maintenance operation on a target device; and A learning step of learning a component to be replaced or repaired or a new component after replacement in the target device based on a data set including a combination of device information of the target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device Among them, The learning step learns a component to be replaced or repaired or a new component after replacement in the target device based on the reward.

14. A maintenance operation assistance method having: A calculation step of calculating a reward based on evaluation information which is information for evaluating the result of a maintenance operation on a target device; and A learning step of learning an operation process of a maintenance operation on the target device based on a data set including a combination of device information of the target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device Among them, The learning step learns an operation process of a maintenance operation on the target device based on the reward.

15. A maintenance operation assistance program for causing a computer to execute: A first acquisition step of acquiring a data set including a combination of device information of a target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device; A second acquisition step of acquiring operation content information in which the content of the maintenance operation performed by a maintenance operator on the target device, namely, a component that has been replaced or repaired or a new component after replacement, is recorded; and A learning step of learning in such a manner that the data set acquired in the first acquisition step is associated with the component that has been replaced or repaired or the new component after replacement recorded in the operation content information acquired in the second acquisition step.

16. A maintenance operation assistance program for causing a computer to execute: A first acquisition step of acquiring a data set including a combination of device information of a target device and operation information of the target device, or a data set including a combination of device information of the target device and phenomenon information indicating a phenomenon related to the target device; A second acquisition step of acquiring operation content information in which the content of the maintenance operation performed by a maintenance operator on the target device, namely, the operation process, is recorded; and A learning step of learning in such a manner that the data set acquired in the first acquisition step is associated with the operation process recorded in the operation content information acquired in the second acquisition step.

17. A maintenance operation assistance program for causing a computer to execute: A calculation step of calculating a reward based on evaluation information which is information for evaluating the result of a maintenance operation on a target device; and A learning step, which learns about components to be replaced or repaired in the target device or new components after replacement based on a data set that combines device information of the target device and operation information of the target device, or a data set that combines device information of the target device and phenomenon information indicating phenomena related to the target device. Among them, The learning step learns about components to be replaced or repaired in the target device or new components after replacement based on the reward.

18. A maintenance operation assistance program for causing a computer to execute: A calculation step of calculating a reward based on evaluation information, which is information for evaluating the result of a maintenance operation for a target device; and A learning step of learning about the operation process of a maintenance operation for the target device based on a data set that combines device information of the target device and operation information of the target device, or a data set that combines device information of the target device and phenomenon information indicating phenomena related to the target device. Among them, The learning step learns about the operation process of a maintenance operation for the target device based on the reward.

Images