JP7325630B2 - Air conditioner diagnostic system and learning device - Google Patents

Description

TECHNICAL FIELD The present disclosure relates to an air conditioner diagnostic system and learning device.

2. Description of the Related Art Diagnosis devices for air conditioners are conventionally known. For example, a self-diagnostic device for an air conditioner control unit described in Patent Document 1 determines failure of a display unit, a remote control unit, etc. by a program command when a self-diagnostic switch is on. In the field, only service company personnel can turn on this self-diagnostic switch.

JP-A-58-178136

However, in Patent Literature 1, the cause of the abnormality, the part that needs to be replaced, or the appropriate worker to perform the repair cannot be specified unless the personnel of the service company visit the site.

SUMMARY OF THE INVENTION Therefore, an object of the present disclosure is to provide an air conditioner diagnostic system and a learning device that can reduce the number of times personnel are dispatched to the site.

An air conditioner diagnostic system of the present disclosure is an air conditioner diagnostic system that remotely diagnoses a plurality of air conditioners. This diagnosis system includes an inference device that infers the cause of an air conditioner abnormality from input data including at least air conditioner model data and air conditioner operation data at the time when an air conditioner abnormality occurred. and a display device for displaying the cause of the abnormality.

The learning device of the present disclosure comprises input data including at least air conditioner model data and air conditioner operation data at the time when an air conditioner abnormality occurred, and teacher data representing the cause of the air conditioner abnormality. A data acquisition unit that acquires data for learning; and a model generation unit that generates a trained model for inferring.

According to the present disclosure, the number of dispatches of personnel to the site can be reduced.

FIG. 3 is a diagram for explaining the flow of maintenance service for a conventional air conditioner; It is a figure showing the structure of the air conditioner 21 of embodiment, and the diagnostic system 1. FIG. 2 is a diagram showing the configuration of a learning device 7; FIG. 3 is a diagram showing an example of model data K1 and K2; FIG. 4 is a diagram showing an example of operating data A1 to An; FIG. FIG. 3 is a diagram showing an example of air conditioner abnormality causes B1 to Bm. 1 is a diagram showing the configuration of a neural network according to Embodiment 1; FIG. 5 is a flow chart showing a procedure of learning by the learning device 7; 1 is a diagram showing the configuration of an inference device 10; FIG. 4 is a flowchart showing the procedure of inference by the inference device 10; FIG. 4 is a diagram showing an example of a replacement part; FIG. 10 is a diagram showing the configuration of a neural network according to Embodiment 2; FIG. FIG. 10 is a diagram showing the configuration of a neural network according to Embodiment 3; FIG. 4 is a diagram showing an example of worker characteristics; FIG. 13 is a diagram showing the configuration of a neural network according to Embodiment 4; FIG. 13 is a diagram showing the configuration of an inference device 10 according to a fourth embodiment; FIG. It is a figure showing worker data. FIG. 13 is a diagram showing the configuration of a neural network according to Embodiment 5; FIG. 14 is a diagram showing the configuration of a neural network according to Embodiment 6; FIG. 12 is a diagram showing the configuration of a neural network according to Embodiment 7; 4 is a diagram showing examples of installation environment data E1 to E3; FIG. FIG. 10 is a diagram representing another example of a worker's characteristics; 2 is a diagram showing the hardware configuration of a learning device 7 or an inference device 10; FIG.

Embodiments will be described below with reference to the drawings.
Embodiment 1.
FIG. 1 is a diagram for explaining the flow of maintenance service for a conventional air conditioner.

The owner owns the air conditioner. The service company performs tasks such as commissioning, inspection, and repair. Service parts are stocked by estimating the number of holdings based on past failure records.

(1) The owner knows that an abnormality has occurred in the air conditioner. (2) The owner requests the operator of the service company for repair. (3) Scheduling for dispatching workers from the service company is made between the owner and the service company. (4) The operator of the service company instructs dispatch of workers. (5) Service company workers check the site. (6) A service company worker diagnoses the cause of an abnormality in the air conditioner. (7) Service company workers arrange parts to be replaced as necessary. (8) The operator of the service company adjusts the schedule for the next repair visit by the worker according to the parts arrangement status. (9) Parts are stocked at the service company. (10) The service company worker visits the site again to perform repairs based on the result of schedule adjustment.

As described above, the conventional maintenance service requires the personnel of the service company to visit the site at least twice before the repair is completed, which is time-consuming and costly. The initial response (first visit) is not likely to be regarded as a repair, and in many cases the owner does not receive the full dispatch fee. Since a quick initial response is required, service companies need to secure personnel on the premise of the busy season. Furthermore, during busy periods, the workload of workers becomes excessive. Depending on the worker, there are strengths and weaknesses in repair content, and there are differences in work time and repair level due to differences in skills. Since there is no clear indicator of how much repair parts should be in stock, it is necessary to stock evenly, resulting in wasted inventory management costs.

FIG. 2 is a diagram showing the configuration of the air conditioner 21 and the diagnostic system 1 for the air conditioner according to the embodiment.

A plurality of air conditioners 21 and diagnostic system 1 are connected by communication line 31 . The diagnostic system 1 may be installed, for example, at a service company.

The air conditioner 21 includes an indoor unit 22 , an outdoor unit 23 , a remote controller 24 and a control device 25 . The control device 25 includes an indoor unit controller 26 , an outdoor unit controller 27 and a system controller 28 .

The indoor unit 22 is installed indoors. The outdoor unit 23 is installed outdoors. The remote controller 24 receives user's operation input. The indoor unit controller 26 controls the indoor unit 22 . The outdoor unit controller 27 controls the outdoor unit 23 . The system controller 28 controls the air conditioner 21 as a whole.

The diagnostic system 1 includes an input device 2 , a communication circuit 3 , a display device 4 , a data storage device 5 , a trained model storage device 6 , a learning device 7 and an inference device 10 .

The input device 2 receives operation input from, for example, an operator of a service company.
The communication circuit 3 exchanges data with the plurality of air conditioners 21 through the communication line 31 .

The data storage device 5 stores the operation data transmitted from the plurality of air conditioners 21 through the communication line 31 with time stamps and device IDs. This device ID is an ID that identifies the air conditioner 21 that has transmitted the operation data. When an abnormality occurs in the air conditioner 21, the data storage device 5 stores the cause of the abnormality diagnosed by a service company worker or the like with a time stamp at the time of occurrence of the abnormality and the device ID. This device ID is an ID that identifies the diagnosed air conditioner.

The learning device 7 generates a trained model for inferring the cause of the abnormality of the air conditioner from at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner.

A trained model storage device 6 stores a trained model generated by the learning device 7 .

The inference device 10 infers the cause of the abnormality of the air conditioner from at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner.

The display device 4 displays the inference result by the inference device 10 .
<Learning phase>
FIG. 3 is a diagram showing the configuration of the learning device 7. As shown in FIG.

The learning device 7 includes a data acquisition section 71 and a model generation section 72 .
Data acquisition unit 71 reads out the combination of the time-stamped cause of abnormality and the device ID stored in data storage device 5 . The data acquisition unit 71 further reads the time-stamped operation data stored in the data storage device 5 in association with the read device ID. The data acquisition unit 71 acquires input data including at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality of the air conditioner occurred, and the cause of the abnormality of the air conditioner, from the read data. Acquisition of learning data consisting of teaching data representing At least the point in time when the abnormality of the air conditioner occurs means the time t1 when the abnormality of the air conditioner occurs and the time within the predetermined time Δt before the time t1. Δt≧0. The model data can be identified from the read device ID.

FIG. 4 is a diagram showing an example of model data K1-K2. The model data includes the model number and product name.

FIG. 5 is a diagram showing an example of the operating data A1-An. The operating data is any data that can represent the operating state of the air conditioner 21 . For example, the operating data includes the operating rate of the compressor, set temperature, outside air temperature, operating mode, compressor discharge temperature, compressor suction temperature, degree of subcooling, compressor frequency, expansion valve opening, indoor unit heat exchanger inlet temperature, indoor unit heat exchanger outlet temperature, outdoor unit heat exchanger inlet temperature, outdoor unit heat exchanger outlet temperature, indoor unit heat exchanger inlet pressure, indoor unit Including the outlet pressure of the heat exchanger of the outdoor unit, the inlet pressure of the heat exchanger of the outdoor unit, the outlet pressure of the outdoor unit heat exchanger, the discharge pressure of the compressor, the suction pressure of the compressor, the degree of heating, etc.

FIG. 6 is a diagram showing an example of air conditioner abnormality causes B1 to Bm. These represent causes of abnormality that can be judged from the operating data of the air conditioner. For example, the cause of an error is an error display on the remote controller, an error in indoor unit communication, an error in discharge temperature, a heating operation failure, an indoor unit stoppage, an error in low pressure, an error in high pressure, an error in drain pump, an error in the dust box, an error in power signal synchronization, Including water leakage, address error, no response error, drain pump, water outage, outside temperature, remote control board, gas leak, drain sensor submerged, filter malfunction, etc.

When the cause of the abnormality of the air conditioner is Bi, the data acquisition unit 71 inputs at least the model data K1 to K2 of the air conditioner and the operation data A1 to An of the air conditioner at the time when the abnormality of the air conditioner occurred. Learning data including data and teacher data Z1 to Zm representing the cause of abnormality Bi of the air conditioner is obtained. However, Zi=1 and Zj=0 (j≠i).

The model generation unit 72 uses the learning data to infer the cause of the abnormality of the air conditioner from at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality of the air conditioner occurred. Generate a trained model.

The model generating unit 72 generates a trained model by so-called supervised learning according to, for example, a neural network model. Supervised learning refers to a method of inferring a result from an input by providing a set of data of factors and results (labels) to the model generation unit 72 to learn features in the learning data.

A neural network consists of an input layer made up of multiple neurons, an intermediate layer (hidden layer) made up of multiple neurons, and an output layer made up of multiple neurons. The intermediate layer may be one layer, or two or more layers. The activation function of the output units of the output layer can be a softmax function.

FIG. 7 is a diagram showing the configuration of a neural network according to Embodiment 1. FIG. FIG. 7 shows a multi-layered neural network.

When multiple inputs (factors) are input to the input layer, the input layer multiplies the value of each input by a weight, and the multiplication result is sent to the next layer. The next layer multiplies the input multiplication result and the weight, and the multiplication result is sent to the next layer. Finally, the output data (result) is output from the output layer. This output result changes depending on the weight.

Data input to the input units of the neural network are model data K1-K2 and operation data A1-An of the air conditioners, and data output from the output units of the output layer are the causes of abnormality B1-Bm.

The neural network performs so-called Learn the cause of anomalies by supervised learning. That is, the neural network inputs model data K1-K2 and operation data A1-An of air conditioners to the input layer, and causes abnormality causes B1-Bm output from the output layer to approach teacher data Z1-Zm. Learn by adjusting the weights.

The model generation unit 72 generates a learned model by executing the learning as described above, and outputs the learned model to the learned model storage device 6 .

The learned model storage device 6 stores the learned model output from the model generation unit 72 .

FIG. 8 is a flow chart showing the procedure of learning by the learning device 7. As shown in FIG.
In step S101, the data acquisition unit 71 acquires learning data including air conditioner model data K1 to K2, operation data A1 to An, and teacher data Z1 to Zm. Although the air conditioner model data K1-K2, the operation data A1-An, and the teacher data Z1-Zm (correct answers) are obtained at the same time, the air conditioner model data K1-K2 and operation data A1-An , and teacher data Z1-Zm may be input in association with each other, and the air conditioner model data K1-K2, operation data A1-An, and teacher data Z1-Zm may be obtained at different timings.

In step S102, the model generation unit 72 creates the air conditioner model data K1 to K2, the operation data A1 to An, and the teacher data Z1 to Zm, which are acquired by the data acquisition unit 71, based on the (correct) combination. A trained model is generated by so-called supervised learning according to the training data obtained.

In step S<b>103 , the learned model storage device 6 stores the learned model generated by the model generation unit 72 .

<Utilization phase>
FIG. 9 is a diagram showing the configuration of the inference device 10. As shown in FIG.

The inference device 10 includes a data acquisition unit 73 and an inference unit 74 .
The data acquisition unit 73 reads the time-stamped operation data stored in the data storage device 5 in association with the device ID of the air conditioner to be inferred. The data acquisition unit 73 acquires input data including at least the air conditioner model data K1 to K2 and the air conditioner operation data A1 to An at the time when the abnormality occurred in the air conditioner from the read data. Here, at least the point in time when the abnormality of the air conditioner occurs means the time t2 when the abnormality of the air conditioner occurs and the time within the predetermined time Δt before the time t2. Δt≧0. The model data K1-K2 can be specified from the device ID of the air conditioner to be inferred.

The inference unit 74 infers the cause of the abnormality of the air conditioner from at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality of the air conditioner occurred, which is stored in the learned model storage device 6. The input data acquired by the data acquisition unit 73 is input to the trained model, and the data representing the cause of the abnormality of the air conditioner is output.

The inference unit 74 uses a trained neural network.
The data input to the input units of the input layer of the neural network are model data K1-K2 and operation data A1-An of the air conditioners, and the data output from the output units of the output layer are the causes of abnormality B1-Bm. is. The inference unit 74 determines the maximum value Bk among the output causes of abnormality B1 to Bm as the cause of the current abnormality of the air conditioner.

FIG. 10 is a flow chart showing the inference procedure by the inference device 10 .
In step S201, the data acquisition unit 73 acquires input data including model data K1 to K2 and operation data A1 to An of the air conditioner.

In step S202, the inference unit 74 inputs input data including air conditioner model data K1 to K2 and operation data A1 to An to the learned model stored in the learned model storage device 6. FIG.

In step S203, the inference unit 74 obtains outputs B1 to Bm of the trained model. When the maximum value among the causes of abnormality B1 to Bm is Bk, the inference unit 74 determines the cause of abnormality Bk as the cause of the current abnormality of the air conditioner.

As described above, according to the present embodiment, when an abnormality occurs in an air conditioner, it is possible to remotely diagnose the cause of the abnormality based on the model data and operation data of the air conditioner. This reduces the number of site visits by service company personnel. As a result, it is possible to suppress the taking-out and improve the cost resistance.

Embodiment 2.
The data storage device 5 stores the operation data transmitted from the plurality of air conditioners 21 through the communication line 31 with time stamps and device IDs. This device ID is an ID that identifies the air conditioner 21 that has transmitted the operation data. The data storage device 5 stores a time stamp at the time when an abnormality occurred in the air conditioner 21 and a replacement part out of a plurality of parts constituting the air conditioner at the time of repair by a worker of a service company or the like. and the device ID are added and stored. This device ID is an ID that identifies the repaired air conditioner.

<Learning phase>
FIG. 11 is a diagram showing examples of replacement parts. These represent replacement parts that can be determined based on the model data and operation data of the air conditioner, or replacement parts that can be determined based on the cause of an abnormality in the air conditioner. For example, replacement parts include control boards, solenoid valves, thermistors, outdoor unit fans, indoor unit fans, low pressure pressure sensors, outdoor unit heat exchangers, indoor unit heat exchangers, drain sensors, drain pipes, float switches, water supply Includes tank, humidification solenoid valve, fuse, motor, compressor, shunt controller, Lossnay, remote controller, and system controller.

Data acquiring portion 71 reads out a combination of a replacement part and a device ID among a plurality of parts constituting the air conditioner with time stamps stored in data storage device 5 . The data acquisition unit 71 further reads the time-stamped operation data stored in the data storage device 5 in association with the read device ID. The data acquisition unit 71 obtains input data including at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality of the air conditioner occurred, and a plurality of data constituting the air conditioner. learning data consisting of training data representing replacement parts among the parts of . Here, at least the point in time when the abnormality of the air conditioner occurs means the time t1 when the abnormality of the air conditioner occurs and the time within the predetermined time Δt before the time t1. Δt≧0. The model data can be identified from the read device ID.

When the replacement part is Ci among the plurality of parts constituting the air conditioner, the data acquisition unit 71 obtains at least the model data K1 to K2 of the air conditioner and the operation of the air conditioner at the time when the abnormality occurred in the air conditioner. Input data including data A1 to An and learning data including teacher data Z1 to Zp representing component parts Ci of the air conditioner are obtained. However, Zi=1 and Zj=0 (j≠i).

The model generating unit 72 uses learning data to infer replacement parts for the air conditioner from at least the model data of the air conditioner and the operation data of the air conditioner at the time when an abnormality occurred in the air conditioner. Generate a trained model. The model generating unit 72 generates a trained model by so-called supervised learning according to, for example, a neural network model.

FIG. 12 is a diagram showing the configuration of a neural network according to Embodiment 2. FIG.
The data input to the input units of the input layer of the neural network are the model data K1-K2 and the operation data A1-An of the air conditioners, and the data output from the output units of the output layer are the replacement parts C1-Cp. is. The activation function of the output units of the output layer can be a softmax function.

The neural network performs so-called Learn replacement parts by supervised learning. That is, the neural network inputs the model data K1 to K2 and the operation data A1 to An of the air conditioner to the input layer, and the replacement parts C1 to Cp output from the output layer are made to approach the teacher data Z1 to Zp. Learn by adjusting the weights.

The model generation unit 72 generates a learned model by executing the learning as described above, and outputs the learned model to the learned model storage device 6 .

<Utilization phase>
The inference device 10 uses input data including at least the air conditioner model data and the air conditioner operation data at the time when the abnormality occurred in the air conditioner to determine a replacement part among a plurality of parts constituting the air conditioner. infer.

The data acquisition unit 73 reads the time-stamped operation data stored in the data storage device 5 in association with the device ID of the air conditioner to be inferred. The data acquisition unit 73 acquires input data including at least the air conditioner model data K1 to K2 and the air conditioner operation data A1 to An at the time when the abnormality occurred in the air conditioner from the read data. Here, at least the point in time when the abnormality of the air conditioner occurs means the time t2 when the abnormality of the air conditioner occurs and the time within the predetermined time Δt before the time t2. Δt≧0. The model data can be identified from the device ID of the air conditioner to be inferred.

The inference unit 74 is a learned model that infers a replacement part among a plurality of parts constituting the air conditioner from at least the model data of the air conditioner at the time when the abnormality occurred in the air conditioner and the operation data of the air conditioner. input data acquired by data acquisition unit 73, and outputs data representing replacement parts among a plurality of parts constituting the air conditioner.

The inference unit 74 uses a trained neural network.
The data input to the input units of the input layer of the neural network are the model data K1-K2 and the operation data A1-An of the air conditioners, and the data output from the output units of the output layer are the replacement parts C1-Cp. is. The inference unit 74 determines the maximum value Ck among the output replacement parts C1 to Cp as the current replacement part for the air conditioner.

As described above, according to the present embodiment, when an abnormality occurs in the air conditioner, replacement parts can be specified remotely based on the model data and operation data of the air conditioner. This reduces the number of site visits by service company personnel. As a result, it is possible to suppress the taking-out and improve the cost resistance.

Embodiment 3.
When an abnormality occurs in the air conditioner 21, the data storage device 5 stores the cause of the abnormality of the air conditioner diagnosed by a service company worker or the like, and the replacement parts among the plurality of parts constituting the air conditioner. are associated with each other and stored.

<Learning phase>
The data acquisition unit 71 determines the cause of the abnormality of the air conditioner based on the cause of the abnormality of the air conditioner stored in association with the data storage device 5 and the replacement part among the plurality of parts constituting the air conditioner. learning data including input data including the input data and teacher data representing replacement parts among a plurality of parts constituting the air conditioner.

When the cause of abnormality of the air conditioner is Bk and the replacement part among the plurality of parts constituting the air conditioner is Ci, the data acquisition unit 71 inputs data including data X1 to Xm representing the cause of abnormality Bk of the air conditioner. Data and teacher data Z1 to Zp representing the replacement parts Ci of the air conditioner are generated. Here, Xk=1 and Xj=0 (j≠k). Zi=1, Zj=0 (j≠i).

The model generation unit 72 uses the learning data to generate a trained model for inferring the replacement parts of the air conditioner from the input data including the cause of the abnormality of the air conditioner.

The model generating unit 72 generates a trained model by so-called supervised learning according to, for example, a neural network model.

FIG. 13 is a diagram showing the configuration of a neural network according to Embodiment 3. FIG.
The data input to the input units of the input layer of the neural network are the data X1 to Xm representing the cause of abnormality Bk of the air conditioner, and the data output from the output units of the output layer are the replacement parts C1 to Cp. . The activation function of the output units of the output layer can be a softmax function.

The neural network performs so-called supervised learning in accordance with learning data created based on a combination of data X1 to Xm and Zi to Zp (teacher data) representing the causes of abnormalities in the air conditioner acquired by the data acquisition unit 71. , to learn replacement parts. That is, the neural network inputs the data X1 to Xm representing the cause of abnormality of the air conditioner to the input layer, and adjusts the weights so that the replacement parts C1 to Cp output from the output layer approach the teacher data Z1 to Zp. Learn by doing.

The model generation unit 72 generates a learned model by executing the learning as described above, and outputs the learned model to the learned model storage device 6 .

<Utilization phase>
When an abnormality occurs in air conditioner 21 , air conditioner 21 transmits data representing the cause of the abnormality to diagnosis system 1 . For example, it is assumed that the air conditioner 21 can identify the cause of an abnormality of the air conditioner from the operating state of the air conditioner 21 within its own device.

When an abnormality occurs in the air conditioner, the inference device 10 of the diagnostic system 1 determines the components of the air conditioner 21 from the input data including the cause of the abnormality of the air conditioner acquired from the air conditioner 21. Inferring our replacement parts.

The data acquisition unit 73 acquires input data including the cause of the abnormality from the air conditioner 21 when an abnormality occurs in the air conditioner 21 .

When the cause of abnormality of the air conditioner is Bk, data acquisition unit 73 generates input data including data X1 to Xm representing cause of abnormality Bk of the air conditioner. Here, Xk=1 and Xj=0 (j≠k).

The inference unit 74 obtains data from the input data including the cause of abnormality of the air conditioner stored in the learned model storage device 6 to the learned model for inferring the replacement part among the plurality of parts constituting the air conditioner. The input data from the air conditioner 21 acquired by the unit 73 is input, and the data representing the replacement part among the plurality of parts constituting the air conditioner is output.

The inference unit 74 uses a trained neural network. The data input to the input units of the input layer of the neural network are data X1 to Xm representing the causes of abnormalities in the air conditioner, and the data output from the output units of the output layer are replacement parts C1 to Cp. where Xk=1, Xj=0 (j≠k).

The inference unit 74 determines the maximum value Ck among the replacement parts C1 to Cp as the current replacement part for the air conditioner.

As described above, according to the present embodiment, when an abnormality occurs in the air conditioner, it is possible to remotely specify a replacement part based on the cause of the abnormality in the air conditioner. This reduces the number of site visits by service company personnel. As a result, it is possible to suppress the taking-out and improve the cost resistance.

Modified example 1 of the third embodiment.
In Embodiment 3, the data acquisition unit 73 acquires input data including the cause of the abnormality from the air conditioner 21 when an abnormality occurs in the air conditioner 21 . The inference unit 74 adds the information from the air conditioner 21 acquired by the data acquisition unit 73 to the trained model for inferring the replacement part among the plurality of parts constituting the air conditioner from the input data including the cause of the abnormality of the air conditioner. Input data is input, and data representing a replacement part among a plurality of parts constituting the air conditioner is output.

In this modification, as in the first embodiment, the inference unit 74 determines the cause of the abnormality of the air conditioner from at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner. The input data acquired by the data acquisition unit 73 is input to the trained model to be inferred, and the data representing the cause of the abnormality of the air conditioner is output.

The inference unit 74 further adds data including the cause of abnormality inferred by the inference unit 74 to a trained model for inferring a replacement part among a plurality of parts constituting the air conditioner from input data including the cause of the abnormality of the air conditioner. is input as input data, and data representing a replacement part among a plurality of parts constituting the air conditioner is output.

Embodiment 4.
The data storage device 5 stores the operation data transmitted from the plurality of air conditioners 21 through the communication line 31 with time stamps and device IDs. This device ID is an ID that identifies the air conditioner 21 that has transmitted the operation data. When an abnormality occurs in the air conditioner 21, the data storage device 5 stores the characteristics of the worker who repaired the air conditioner with a time stamp at the time when the abnormality occurred and the device ID. This device ID is an ID that identifies the repaired air conditioner.

FIG. 14 is a diagram illustrating an example of worker characteristics. These are the characteristics of a worker who repairs an air conditioner that can be judged from the model data and operation data of the air conditioner, the characteristics of the worker that repairs the air conditioner that can be judged from the cause of an abnormality in the air conditioner, Characteristics of workers who repair air conditioners that can be judged from replacement parts of air conditioners, or workers who repair air conditioners that can be judged from the cause of abnormality of air conditioners and replacement parts of air conditioners characterize. For example, the worker characteristics D1, D2, D3, D4, D5, D6, and D7 are, respectively, years of experience, skill level regarding outdoor unit, technical level regarding indoor unit, technical level regarding electronic control, technical level regarding machine, communication technical level related to display, and technical level related to display. Each value of D1 to Ds can be 0 or more and 1 or less.

<Learning phase>
Data acquisition unit 71 reads out the time-stamped combination of the characteristics of the worker who repaired the air conditioner and the device ID stored in data storage device 5 . The data acquisition unit 71 further reads the time-stamped operation data stored in the data storage device 5 in association with the read device ID.

The data acquisition unit 71 obtains, from the read data, input data including at least the air conditioner model data K1 to K2 and the air conditioner operation data A1 to An at the time when the air conditioner abnormality occurred, and the air conditioner Learning data including teaching data Z1-Zs representing characteristics D1-Ds of the worker who repaired the harmony machine is obtained. Here, at least the point in time when the abnormality of the air conditioner occurs means the time t1 when the abnormality of the air conditioner occurs and the time within the predetermined time Δt before the time t1. Δt≧0. The model data can be identified from the read device ID.

The model generation unit 72 uses the learning data to determine at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner. Generate a trained model for inferring traits. The model generating unit 72 generates a trained model by so-called supervised learning according to, for example, a neural network model.

FIG. 15 is a diagram showing the configuration of a neural network according to Embodiment 4. FIG.
The data input to the input units of the input layer of the neural network are the model data K1 to K2 and the operation data A1 to An of the air conditioners, and the data output from the output units of the output layer are the repair of the air conditioner. are the characteristics D1 to Ds of the workers who performed the

The neural network operates according to learning data created based on a combination of air conditioner model data K1-K2, operation data A1-An, and Zi-Zs (teacher data) acquired by the data acquisition unit 71. The characteristics of workers who repaired air conditioners are learned by supervised learning. That is, in the neural network, the model data K1 to K2 and the operation data A1 to An of the air conditioner are input to the input layer, and the worker characteristics D1 to Ds output from the output layer approach the teacher data Z1 to Zs. It learns by adjusting the weights as follows.

The model generation unit 72 generates a learned model by executing the learning as described above, and outputs the learned model to the learned model storage device 6 .

<Utilization phase>
FIG. 16 is a diagram showing the configuration of inference apparatus 10 according to the fourth embodiment.

The inference device 10 infers the characteristics of the worker who repaired the air conditioner from input data including at least air conditioner model data and air conditioner operation data at the time when an abnormality occurred in the air conditioner. .

The inference device 10 includes a data acquisition unit 73 , an inference unit 74 , and an operator selection unit 81 .
The data acquisition unit 73 reads the time-stamped operation data stored in the data storage device 5 in association with the device ID of the air conditioner to be inferred. The data acquisition unit 73 acquires input data including at least the air conditioner model data K1 to K2 and the air conditioner operation data A1 to An at the time when the abnormality occurred in the air conditioner from the read data. Here, at least the point in time when the abnormality of the air conditioner occurs means the time t2 when the abnormality of the air conditioner occurs and the time within the predetermined time Δt before the time t2. Δt≧0. The model data can be identified from the device ID of the air conditioner to be inferred.

The inference unit 74 has learned to infer the characteristics of the worker who repairs the air conditioner from the input data including the air conditioner model data and the air conditioner operation data at least at the time when the abnormality occurred in the air conditioner. The input data acquired by the data acquisition unit 73 is input to the model, and the data representing the characteristics of the worker who repairs the air conditioner is output.

The inference unit 74 uses a trained neural network stored in the trained model storage device 6 . The data input to the input units of the input layer of the neural network are the model data K1-K2 and the operation data A1-An of the air conditioners, and the data output from the output units of the output layer are the repair of the air conditioner. Data D1 to Ds represent the characteristics of the worker who performs the work.

The worker selection unit 81 selects a worker to be dispatched for repair according to the inferred characteristics of the worker and the characteristics and schedules of the plurality of workers.

FIG. 17 is a diagram representing worker data.
Worker data defines characteristics and dispatch availability dates for each worker.

For example, the characteristics D1, D2, D3, D4, D5, D6, D7, . , 0.2 . Worker M1 can be dispatched from 7/14 to 7/29.

The worker selection unit 81 calculates the square sum Su of the differences between the worker characteristics D1 to Ds estimated by the inference unit 74 and the characteristics D1 to Ds of each worker Mu. The worker selection unit 81 identifies the worker with the smallest Su, and if the dispatchable date of the worker is included in the on-site acceptable schedule, the worker is selected as a worker to be dispatched for repair. certify. If the dispatchable date of the worker is not included in the schedule acceptable at the site, the worker selection unit 81 identifies the worker whose Su is the next smallest, and the worker's dispatchable date is If it is included in the acceptable schedule, the worker is certified as a worker to be dispatched for repair. If the worker's dispatchable date is not included in the site-acceptable schedule, the above processing is repeated until another worker's dispatchable date is included in the site-acceptable schedule.

As described above, according to the present embodiment, when an abnormality occurs in an air conditioner, a suitable worker who remotely repairs the air conditioner is selected based on the model data and operation data of the air conditioner. can be specified. This reduces the number of site visits by service company personnel. In addition, the trainee can gain experience by combining with the right worker and having the trainee accompany the worker. Furthermore, by providing training on skills that are lacking according to engineer data, it is possible to raise the technical capabilities of the service company.

Embodiment 5.
When an abnormality occurs in the air conditioner 21, the data storage device 5 stores the cause of the abnormality of the air conditioner diagnosed by the worker of the service company, etc., and the characteristics of the worker who repaired the air conditioner. They are associated and stored.

<Learning phase>
The data acquisition unit 71 obtains input data including the cause of the abnormality of the air conditioner based on the cause of the abnormality of the air conditioner stored in association with the data storage device 5 and the characteristics of the worker who repaired the air conditioner. and training data representing the characteristics of the worker who repaired the air conditioner.

When the cause of abnormality of the air conditioner is Bk, the data acquisition unit 71 obtains input data including data X1 to Xm representing the cause of abnormality Bk of the air conditioner and characteristic D1 of the worker who repaired the air conditioner. Learning data including teacher data Z1-Zs representing .about.Ds is obtained. Here, Xk=1 and Xj=0 (j≠k).

The model generation unit 72 uses the learning data to generate a trained model for inferring the characteristics of the worker who repaired the air conditioner from the cause of the abnormality of the air conditioner. The model generating unit 72 generates a trained model by so-called supervised learning according to, for example, a neural network model.

18 is a diagram showing the configuration of a neural network according to Embodiment 5. FIG.
The data input to the input units of the input layer of the neural network are data X1 to Xm representing the cause of abnormality Bk, and the data output from the output units of the output layer are the data of the worker who repaired the air conditioner. These are characteristics D1 to Ds.

The neural network performs so-called supervised learning according to learning data created based on a combination of data X1 to Xm representing the cause of abnormality acquired by the data acquisition unit 71 and Zi to Zs (teacher data). Learn the characteristics of the worker who repaired the machine. That is, the neural network inputs the data X1 to Xm representing the cause of abnormality to the input layer, and adjusts the weights so that the worker characteristics D1 to Ds output from the output layer approach the teacher data Z1 to Zs. learn in

The model generation unit 72 generates a learned model by executing the learning as described above, and outputs the learned model to the learned model storage device 6 .

<Utilization phase>
When an abnormality occurs in air conditioner 21 , air conditioner 21 transmits data representing the cause of the abnormality to diagnosis system 1 . For example, it is assumed that the air conditioner 21 can identify the cause of an abnormality of the air conditioner from the operating state of the air conditioner 21 within its own device.

When an abnormality occurs in the air conditioner, the inference device 10 of the diagnostic system 1 uses the input data including the cause of the abnormality of the air conditioner acquired from the air conditioner 21 to determine the characteristics of the worker who repairs the air conditioner. to infer

The data acquisition unit 73 acquires input data including the cause of the abnormality from the air conditioner 21 when an abnormality occurs in the air conditioner 21 .

When the cause of abnormality of the air conditioner is Bk, data acquisition unit 73 generates input data including data X1 to Xm representing cause of abnormality Bk of the air conditioner. Here, Xk=1 and Xj=0 (j≠k).

The inference unit 74 converts the data acquisition unit 73 into a learned model for inferring the characteristics of the worker who repairs the air conditioner from the input data including the cause of abnormality of the air conditioner stored in the learned model storage device 6. The acquired input data from the air conditioner 21 is input, and data representing the characteristics of the worker of the air conditioner is output.

The inference unit 74 uses a trained neural network. The data input to the input units of the input layer of the neural network are data X1 to Xm representing the cause of abnormality Bk of the air conditioner, and the data output from the output units of the output layer are the work of repairing the air conditioner. data D1 to Ds representing characteristics of a person.

As described above, according to the present embodiment, when an abnormality occurs in an air conditioner, it is possible to identify a qualified worker who remotely repairs the air conditioner based on the cause of the abnormality in the air conditioner. can be done.

Modification 1 of the fifth embodiment.
In Embodiment 5, the data acquisition unit 73 acquires input data including the cause of the abnormality from the air conditioner 21 when an abnormality occurs in the air conditioner 21 . The inference unit 74 applies the input data from the air conditioner 21 acquired by the data acquisition unit 73 to the trained model for inferring the characteristics of the worker who repairs the air conditioner from the input data including the cause of abnormality of the air conditioner. It inputs and outputs data representing the characteristics of the worker who repairs the air conditioner.

In this modification, as in the first embodiment, the inference unit 74 determines the cause of the abnormality of the air conditioner from at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner. The input data acquired by the data acquisition unit 73 is input to the trained model to be inferred, and the data representing the cause of the abnormality of the air conditioner is output.

The inference unit 74 further inputs the data including the cause of abnormality inferred by the inference unit 74 into the trained model for inferring the characteristics of the worker who repairs the air conditioner from the input data including the cause of the abnormality of the air conditioner. , and outputs data representing the characteristics of the worker who repairs the air conditioner.

Embodiment 6.
The data storage device 5 associates, when an abnormality occurs in the air conditioner 21, a replacement part among a plurality of parts constituting the air conditioner with the characteristics of the worker who repaired the air conditioner. memorize.

<Learning phase>
The data acquisition unit 71 obtains an air conditioner based on the characteristics of the operator who repaired the air conditioner and the replacement part among the plurality of parts constituting the air conditioner stored in association with the data storage device 5 . Learning data consisting of input data including replacement parts among a plurality of parts constituting the air conditioner and teacher data representing the characteristics of the worker who repaired the air conditioner is obtained.

When the replacement part among the plurality of parts constituting the air conditioner is Ck, the data acquisition unit 71 obtains input data including data Y1 to Yp representing the replacement part Ck of the air conditioner and the repair data of the air conditioner. Learning data including teaching data Z1 to Zs representing the characteristics D1 to Ds of the worker who performed the training are acquired. Here, Yk=1 and Yj=0 (j≠k).

The model generation unit 72 uses the learning data to create a trained model for inferring the characteristics of the worker who repaired the air conditioner from the replacement parts among the plurality of parts that make up the air conditioner. Generate. The model generating unit 72 generates a trained model by so-called supervised learning according to, for example, a neural network model.

FIG. 19 is a diagram showing the configuration of a neural network according to Embodiment 6. FIG.
The data input to the input units of the input layer of the neural network are data Y1 to Yp representing the replacement parts Ck among the plurality of parts constituting the air conditioner, and the data output from the output units of the output layer are data. , characteristics D1 to Ds of the worker who repaired the air conditioner.

The neural network learns based on a combination of data Y1 to Yp representing replacement parts among a plurality of parts constituting the air conditioner acquired by the data acquisition unit 71 and Zi to Zs (teacher data). According to the data, the characteristics of the worker who repaired the air conditioner are learned by so-called supervised learning. That is, in the neural network, data Y1 to Yp representing replacement parts among a plurality of parts constituting the air conditioner are input to the input layer, and the worker characteristics D1 to Ds output from the output layer are teacher data. Learning is performed by adjusting the weights so as to approach Z1 to Zs.

The model generation unit 72 generates a learned model by executing the learning as described above, and outputs the learned model to the learned model storage device 6 .

<Utilization phase>
When an abnormality occurs in air conditioner 21 , air conditioner 21 transmits to diagnosis system 1 data representing replacement parts among a plurality of parts constituting the air conditioner. For example, it is assumed that the air conditioner 21 can specify a replacement part among a plurality of parts constituting the air conditioner from the operating state or abnormal state of the air conditioner 21 within its own device.

When an abnormality occurs in the air conditioner, the inference device 10 of the diagnostic system 1 uses input data including a replacement part among a plurality of parts constituting the air conditioner acquired from the air conditioner 21 to determine the air conditioner Infer the characteristics of the worker who performs the repair.

The data acquisition unit 73 acquires input data including replacement parts for the air conditioner from the air conditioner 21 when an abnormality occurs in the air conditioner 21 .

The data acquisition unit 73 acquires input data including the replacement part Ck among the plurality of parts forming the air conditioner. When the replacement part for the air conditioner is Ck, the data acquisition unit 73 generates input data including data Y1 to Yp representing the replacement part for the air conditioner Ck. Here, Yk=1 and Yj=0 (j≠k).

The inference unit 74 converts the data acquisition unit 73 into a learned model for inferring the characteristics of the worker who repairs the air conditioner from the input data including the replacement parts of the air conditioner stored in the learned model storage device 6. The acquired input data is input, and data representing the characteristics of the air conditioner operator is output.

The inference unit 74 uses a trained neural network. The data input to the input units of the input layer of the neural network are data Y1 to Yp representing the replacement parts Ck of the air conditioner, and the data output from the output units of the output layer are the work of repairing the air conditioner. data D1 to Ds representing characteristics of a person.

As described above, according to the present embodiment, when an abnormality occurs in the air conditioner, it is possible to identify a qualified worker who remotely repairs the air conditioner based on the replacement parts of the air conditioner. can be done.

Modification of Embodiment 6.
In Embodiment 6, the data acquisition unit 73 acquires input data including replacement parts among a plurality of parts constituting the air conditioner from the air conditioner 21 when an abnormality occurs in the air conditioner 21. . The inference unit 74 adds the air conditioner acquired by the data acquisition unit 73 to a trained model for inferring the characteristics of a worker who repairs the air conditioner from the input data including replacement parts among the plurality of parts constituting the air conditioner. It receives input data from the air conditioner 21 and outputs data representing the characteristics of the worker who repairs the air conditioner.

In this modification, as in Embodiment 2, the inference unit 74 uses at least the model data of the air conditioner and the operation data of the air conditioner at the time when an abnormality occurred in the air conditioner. The input data acquired by the data acquiring unit 73 is input to the trained model for inferring the replacement part among the parts, and the data representing the replacement part among the plurality of parts constituting the air conditioner is output. Alternatively, as in Embodiment 3, the inference unit 74 uses the data acquired by the data acquisition unit 73 as a learned model for inferring the replacement part among the plurality of parts that make up the air conditioner from the cause of an abnormality in the air conditioner. Input data is input, and data representing a replacement part among a plurality of parts constituting the air conditioner is output.

The inference unit 74 further infers the learned model for inferring the characteristics of the worker who repairs the air conditioner from the input data including the replacement part among the plurality of parts constituting the air conditioner. Data including replacement parts among a plurality of parts constituting the air conditioner is input as input data, and data representing the characteristics of the worker who repairs the air conditioner is output.

Embodiment 7.
The data storage device 5 stores, when an abnormality occurs in the air conditioner 21, the cause of the abnormality of the air conditioner diagnosed by a service company worker or the like, and the replacement parts among the plurality of parts constituting the air conditioner. , and the characteristics of the worker who repaired the air conditioner are stored in association with each other.

<Learning phase>
The data acquisition unit 71 acquires the cause of the abnormality of the air conditioner stored in the data storage device 5, the replacement part among the plurality of parts constituting the air conditioner, and the work done to repair the air conditioner. From the characteristics of the worker, learning data consisting of input data including the cause of the abnormality of the air conditioner and replacement parts and teacher data representing the characteristics of the worker who repaired the air conditioner is acquired.

When the cause of abnormality of the air conditioner is Bk and the replacement part of the air conditioner is Cr, the data acquisition unit 71 acquires data X1 to Xm representing the cause of abnormality of the air conditioner and the replacement part Cr of the air conditioner. and training data Z1-Zs representing the characteristics D1-Ds of the workers who repaired the air conditioner are acquired. Here, Xk=1 and Xj=0 (j≠k). Yr=1 and Yj=0 (j≠r).

The model generation unit 72 uses the learning data to generate a trained model for inferring the characteristics of the worker who repaired the air conditioner from the cause of the abnormality of the air conditioner and the replacement parts of the air conditioner. do. The model generating unit 72 generates a trained model by so-called supervised learning according to, for example, a neural network model.

FIG. 20 is a diagram showing the configuration of a neural network according to Embodiment 7. FIG.
The data input to the input unit of the input layer of the neural network are the data X1 to Xm representing the cause of abnormality Bk of the air conditioner and the data Y1 to Yp representing the replacement part Cr of the air conditioner, and the output of the output layer. The data output from the unit are the characteristics D1 to Ds of the worker who repaired the air conditioner.

The neural network follows learning data created based on a combination of data X1 to Xm representing the cause of abnormality and data Y1 to Yp representing replacement parts acquired by the data acquisition unit 71, and Zi to Zs (teacher data). , through so-called supervised learning, the characteristics of the worker who repaired the air conditioner are learned. That is, the neural network inputs data X1 to Xm representing the cause of abnormality of the air conditioner and data Y1 to Yp representing the replacement parts of the air conditioner to the input layer, and outputs the worker characteristics D1 to D1 to Yp from the output layer. Learning is performed by adjusting weights so that Ds approaches teacher data Z1 to Zs.

The model generation unit 72 generates a learned model by executing the learning as described above, and outputs the learned model to the learned model storage device 6 .

<Utilization phase>
When an abnormality occurs in air conditioner 21, air conditioner 21 transmits to diagnostic system 1 data indicating the cause of the abnormality in the air conditioner and replacement parts among a plurality of parts constituting the air conditioner. For example, the air conditioner 21 identifies the cause of an abnormality of the air conditioner and a replacement part among a plurality of parts constituting the air conditioner from the operating state or abnormal state of the air conditioner 21 within its own device. It shall be possible.

The reasoning device 10 of the diagnostic system 1 includes the cause of the abnormality of the air conditioner acquired from the air conditioner 21 and the replacement part among the plurality of parts constituting the air conditioner when the abnormality occurs in the air conditioner. From the input data, infer the characteristics of the worker who repairs the air conditioner.

When an abnormality occurs in the air conditioner 21, the data acquisition unit 73 acquires from the air conditioner 21 input data including the cause of the abnormality of the air conditioner and replacement parts for the air conditioner.

The data acquisition unit 73 acquires input data including an air conditioner abnormality cause Bk and an air conditioner replacement part Cr. When the cause of abnormality of the air conditioner is Bk and the replacement part of the air conditioner is Cr, data acquisition unit 73 acquires data X1 to Xm representing the cause of abnormality Bk of the air conditioner and the replacement part of the air conditioner Cr. Input data is generated that includes representative data Y1-Yp. Here, Xk=1, Xj=0 (j≠k), Yr=1, Yj=0 (j≠r).

The inference unit 74 infers the characteristics of the worker who repairs the air conditioner from the input data including the cause of the abnormality of the air conditioner and the replacement parts of the air conditioner stored in the learned model storage device 6. The input data from the air conditioner 21 acquired by the data acquisition unit 73 is input to the model, and the data representing the characteristics of the worker of the air conditioner is output.

The inference unit 74 uses a trained neural network. The data input to the input unit of the input layer of the neural network are the data X1 to Xm representing the cause of abnormality Bk of the air conditioner and the data Y1 to Yp representing the replacement part Cr of the air conditioner, and the output of the output layer. The data output from the unit are data D1 to Ds representing the characteristics of the worker who repairs the air conditioner.

As described above, according to the present embodiment, when an abnormality occurs in the air conditioner, the air conditioner is repaired remotely based on the cause of the abnormality and replacement parts of the air conditioner. Qualified workers can be identified.

Modification 1 of the seventh embodiment.
In Embodiment 7, when an abnormality occurs in the air conditioner 21, the data acquisition unit 73 obtains from the air conditioner 21 the cause of the abnormality of the air conditioner and the replacement parts among the plurality of parts constituting the air conditioner. Get input data containing The inference unit 74 adds data to a trained model for inferring the characteristics of a worker who repairs an air conditioner from input data including the cause of an abnormality in the air conditioner and replacement parts among a plurality of parts constituting the air conditioner. The input data from the air conditioner 21 acquired by the acquisition unit 73 is input, and the data representing the characteristics of the worker who repairs the air conditioner is output.

In this modified example, as in the modified example of Embodiment 3, the inference unit 74 determines the air conditioner model data and the air conditioner operation data at least at the time when the air conditioner abnormality occurs. The input data acquired by the data acquiring unit 73 is input to the learned model for inferring the cause of abnormality, and the data representing the cause of the abnormality of the air conditioner is output. The inference unit 74 further inputs the cause of the abnormality of the air conditioner estimated by the inference unit 74 to the trained model for inferring the replacement part among the plurality of parts constituting the air conditioner from the cause of the abnormality of the air conditioner. and outputs data representing replacement parts among a plurality of parts constituting the air conditioner.

The inference unit 74 further provides a trained model for inferring the characteristics of the worker who repairs the air conditioner from the input data including the cause of the abnormality of the air conditioner and the replacement parts among the plurality of parts constituting the air conditioner. Characteristic of a worker who repairs an air conditioner by inputting data including the cause of abnormality of the air conditioner inferred by the inference unit 74 and replacement parts among a plurality of parts constituting the air conditioner as input data outputs data representing

Modification.
(1) Although the learning device and the reasoning device are provided inside the diagnostic system, they are not limited to this. The learning device and reasoning device may be provided external to the diagnostic system.

(2) As a learning algorithm used in the model generation unit, deep learning that learns to extract the feature amount itself can be used, and other known methods such as genetic programming, functional logic programming, support vector machines, etc. Machine learning may be performed according to

(3) In the above several embodiments, model data of the air conditioner is input to the learning device and the inference device, but installation environment data may also be input.

FIG. 21 is a diagram showing examples of installation environment data E1 to E3. As shown in FIG. 21, the installation environment data E1 to E3 include the type of building in which the air conditioner is installed, the region in which the air conditioner is installed, and the season.

For example, the installation environment data E1 to E3 are added to the data input to the input layer of the neural network in FIG. 7, and the learning device 7 and the reasoning device 10 can operate as follows. .

When the cause of the abnormality of the air conditioner is Bi, the data acquisition unit 71 of the learning device 7 acquires at least the model data K1 to K2 of the air conditioner and the installation environment data E1 of the air conditioner at the time when the abnormality of the air conditioner occurred. . However, Zi=1 and Zj=0 (j≠i).

The model generating unit 72 of the learning device 7 uses the learning data to obtain data from at least the model data of the air conditioner, the installation environment data of the air conditioner, and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner. , generate a trained model for inferring the cause of anomalies in the air conditioner.

The data acquisition unit 73 of the inference device 10 acquires at least the air conditioner model data K1 to K2, the air conditioner installation environment data E1 to E3, and the air conditioner operation data A1 to Get input data containing An. Here, at least the point in time when the abnormality of the air conditioner occurs means the time t2 when the abnormality of the air conditioner occurs and the time within the predetermined time Δt before the time t2. Δt≧0.

The inference unit 74 of the inference device 10 uses at least the model data of the air conditioner, the installation environment data of the air conditioner, and the data of the installation environment of the air conditioner stored in the learned model storage device 6 at the time when the abnormality occurred in the air conditioner. The input data acquired by the data acquiring unit 73 is input to a trained model for inferring the cause of an abnormality of the air conditioner from the operation data of (1), and the data representing the cause of the abnormality of the air conditioner is output.

For example, the installation environment data E1 to E3 are added to the data input to the input layer of the neural network in FIG. 12, and the learning device 7 and the reasoning device 10 can operate as follows. .

The data acquisition unit 71 of the learning device 7 acquires at least the model data K1 to K2 of the air conditioner at the time when the abnormality occurred in the air conditioner, the air Input data including installation environment data E1 to E3 of the air conditioner and operation data A1 to An of the air conditioner, and learning data including teacher data Z1 to Zp representing the components Ci of the air conditioner are obtained. However, Zi=1 and Zj=0 (j≠i).

The model generation unit 72 of the learning device 7 uses the learning data to obtain at least the model data K1 to K2 of the air conditioner, the installation environment data E1 to E3 of the air conditioner, and the air Generate a trained model for inferring replacement parts for the air conditioner from the operating data of the air conditioner.

The data acquisition unit 73 of the inference device 10 acquires at least the air conditioner model data K1 to K2, the air conditioner installation environment data E1 to E3, and the air conditioner operation data A1 at the time when the air conditioner abnormality occurred. Get the input data containing ~An. Here, at least the point in time when the abnormality of the air conditioner occurs means the time t2 when the abnormality of the air conditioner occurs and the time within the predetermined time Δt before the time t2. Δt≧0.

The inference unit 74 of the inference device 10 uses at least the model data of the air conditioner, the installation environment data of the air conditioner, and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner. The input data acquired by the data acquiring unit 73 is input to a learned model that infers the replacement part among the parts, and the data representing the replacement part among the plurality of parts constituting the air conditioner is output.

For example, installation environment data E1 to E3 are added to the data input to the input layer of the neural network in FIG. 15, and learning device 7 and reasoning device 10 can operate as follows. .

The data acquisition unit 71 of the learning device 7 acquires at least the air conditioner model data K1 to K2, the air conditioner installation environment data E1 to E3, and the air conditioner operation data A1 to Learning data including input data including An and teacher data Z1-Zs representing characteristics D1-Ds of workers who have repaired the air conditioner are obtained. Here, at least the point in time when the abnormality of the air conditioner occurs means the time t1 when the abnormality of the air conditioner occurs and the time within the predetermined time Δt before the time t1. Δt≧0.

The model generating unit 72 of the learning device 7 uses the learning data to obtain data from at least the model data of the air conditioner, the installation environment data of the air conditioner, and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner. , generate a trained model for inferring the characteristics of the worker who repaired the air conditioner.

The data acquisition unit 73 of the inference device 10 acquires at least the air conditioner model data K1 to K2, the air conditioner installation environment data E1 to E3, and the air conditioner operation data A1 to Get input data containing An. Here, at least the point in time when the abnormality of the air conditioner occurs means the time t2 when the abnormality of the air conditioner occurs and the time within the predetermined time Δt before the time t2. Δt≧0.

The inference unit 74 of the inference device 10 determines the type of the air conditioner from input data including at least the model data of the air conditioner, the installation environment data of the air conditioner, and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner. The input data acquired by the data acquisition unit 73 is input to a learned model for inferring the characteristics of the worker who performs the repair, and the data representing the characteristics of the worker who repairs the air conditioner is output.

(4) FIG. 22 is a diagram showing another example of worker characteristics. The characteristics of FIG. 22 include, in addition to the characteristics of FIG. 14, technology levels related to air conditioner models such as technology level D8 related to model A and technology level D9 related to model B. FIG. The values of D8 and D9 can be greater than or equal to 0 and less than or equal to 1.

In some of the above embodiments, the worker characteristic levels of FIG. 22 may be used in place of the worker characteristic levels of FIG.

(5) For example, even if at least one of model data K1-K2 and installation environment data E1-E3 is added to the data input to the input layer of the neural networks of FIGS. good.

(6) The inference device and the learning device described in the above embodiments can be configured with digital circuit hardware or software for corresponding operations. When the functions of the inference device and learning device are implemented using software, the inference device and learning device include, for example, a processor 5002 and a memory 5001 connected by a bus 5003, as shown in FIG. , a program stored in the memory 5001 can be executed by the processor 5002 .

(7) In the above embodiments, the inference device uses the learned model to determine the cause of an abnormality in the air conditioner, replacement parts for the air conditioner, or repair of the air conditioner from the input data acquired by the data acquisition unit. Although it is assumed that the characteristics of the worker who performs the work are inferred, it is not limited to this.

For example, based on rule-based inference or case-based inference, the inference device uses the input data acquired by the data acquisition unit to determine the cause of an abnormality in the air conditioner, replace parts for the air conditioner, or repair the air conditioner. It is also possible to infer the characteristics of the worker.

It should be considered that the embodiments disclosed this time are illustrative in all respects and not restrictive. The scope of the present disclosure is indicated by the scope of claims rather than the above description, and is intended to include all changes within the meaning and scope of equivalence to the scope of claims.

1 diagnostic system, 2 input device, 3 communication circuit, 4 display device, 5 data storage device, 6 learned model storage device, 7 learning device, 10 reasoning device, 21 air conditioner, 22 indoor unit, 23 outdoor unit, 24 Remote controller, 25 control device, 26 indoor unit controller, 27 outdoor unit controller, 28 system controller, l31 communication line, 71, 73 data acquisition unit, 72 model generation unit, 74 inference unit, 81 operator selection unit, 5001 memory, 5002 Processor, 5003 Bus.

Claims (14)

An air conditioner diagnostic system for remotely diagnosing a plurality of air conditioners,
an inference device for inferring characteristics of a worker who repairs the air conditioner from input data including model data of the air conditioner and operation data of the air conditioner at least at the time when an abnormality occurred in the air conditioner; ,
a display device that displays the inferred characteristics of the worker;
The reasoning device is
a data acquisition unit that acquires the input data;
The learned model by deep learning that infers the characteristics of the worker who repairs the air conditioner from at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner. an inference unit that inputs the input data acquired by the data acquisition unit and infers the characteristics of the worker who repairs the air conditioner;
The characteristics of the worker include years of experience, skill level related to the outdoor unit of the air conditioner, technical level related to the indoor unit of the air conditioner, technical level related to electronic control, technical level related to machinery, technical level related to communication, and display A diagnosis system for an air conditioner, including a technical level and a technical level for each model of the air conditioner. An air conditioner diagnostic system for remotely diagnosing a plurality of air conditioners,
an inference device for inferring characteristics of a worker who repairs the air conditioner from input data including causes of abnormalities in the air conditioner;
a display device that displays the inferred characteristics of the worker;
The reasoning device is
a data acquisition unit that acquires the input data;
Repairing the air conditioner by inputting the input data acquired by the data acquisition unit into a trained model by deep learning that infers the characteristics of the worker who repairs the air conditioner from the cause of the abnormality of the air conditioner. and an inference unit that infers the characteristics of the worker who performs
The characteristics of the worker include years of experience, skill level related to the outdoor unit of the air conditioner, technical level related to the indoor unit of the air conditioner, technical level related to electronic control, technical level related to machinery, technical level related to communication, and display A diagnosis system for an air conditioner, including a technical level and a technical level for each model of the air conditioner. An air conditioner diagnostic system for remotely diagnosing a plurality of air conditioners,
an inference device for inferring the cause of an abnormality of the air conditioner from input data including at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner;
a display device that displays the inferred cause of abnormality,
The reasoning device is
a data acquisition unit that acquires the input data;
The data acquisition unit acquires a learned model by deep learning that infers the cause of the abnormality of the air conditioner from at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality of the air conditioner occurs. an inference unit that inputs the input data obtained and infers the cause of the abnormality of the air conditioner,
The inference unit further adds data including the inferred cause of abnormality of the air conditioner to a trained model by deep learning for inferring the characteristics of a worker who repairs the air conditioner from the cause of abnormality of the air conditioner. input as the input data to infer the characteristics of the worker who repairs the air conditioner ;
The characteristics of the worker include years of experience, skill level related to the outdoor unit of the air conditioner, technical level related to the indoor unit of the air conditioner, technical level related to electronic control, technical level related to machinery, technical level related to communication, and display A diagnosis system for an air conditioner, including a technical level and a technical level for each model of the air conditioner. An air conditioner diagnostic system for remotely diagnosing a plurality of air conditioners,
an inference device for inferring characteristics of a worker who repairs the air conditioner from input data including replacement parts among a plurality of parts constituting the air conditioner;
a display device that displays the inferred characteristics of the worker ;
The reasoning device is
a data acquisition unit that acquires the input data;
Inputting the input data acquired by the data acquiring unit to a trained model by deep learning that infers characteristics of a worker who repairs the air conditioner from a replacement part among a plurality of parts constituting the air conditioner. and an inference unit that infers the characteristics of a worker who repairs the air conditioner ,
The characteristics of the worker include years of experience, skill level related to the outdoor unit of the air conditioner, technical level related to the indoor unit of the air conditioner, technical level related to electronic control, technical level related to machinery, technical level related to communication, and display A diagnosis system for an air conditioner, including a technical level and a technical level for each model of the air conditioner. An air conditioner diagnostic system for remotely diagnosing a plurality of air conditioners,
A replacement part among a plurality of parts constituting the air conditioner is inferred from input data including at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner. an inference device;
a display device that displays the inferred replacement part,
The reasoning device is
a data acquisition unit that acquires the input data;
Learning by deep learning for inferring a replacement part among a plurality of parts constituting the air conditioner from at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner. an inference unit that inputs the input data acquired by the data acquisition unit into the model and infers a replacement part among the plurality of parts that constitute the air conditioner;
The inference unit further adds the inferred air conditioner to a trained model by deep learning that infers characteristics of a worker who repairs the air conditioner from replacement parts among a plurality of parts constituting the air conditioner. inputting data including replacement parts among a plurality of parts constituting the air conditioner as the input data, inferring the characteristics of the worker who repairs the air conditioner ;
The characteristics of the worker include years of experience, skill level related to the outdoor unit of the air conditioner, technical level related to the indoor unit of the air conditioner, technical level related to electronic control, technical level related to machinery, technical level related to communication, and display A diagnosis system for an air conditioner, including a technical level and a technical level for each model of the air conditioner. An air conditioner diagnostic system for remotely diagnosing a plurality of air conditioners,
an inference device for inferring the cause of an abnormality of the air conditioner from input data including at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner;
a display device that displays the inferred cause of abnormality,
The reasoning device is
a data acquisition unit that acquires the input data;
The data acquisition unit acquires a learned model by deep learning that infers the cause of the abnormality of the air conditioner from at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality of the air conditioner occurs. an inference unit that inputs the input data obtained and infers the cause of the abnormality of the air conditioner,
The inferring unit further adds the inferred cause of the abnormality of the air conditioner to a trained model by deep learning that infers a replacement part among a plurality of parts constituting the air conditioner from the cause of the abnormality of the air conditioner. by inputting the data including
The inference unit further adds the inferred air conditioner to a trained model by deep learning that infers characteristics of a worker who repairs the air conditioner from replacement parts among a plurality of parts constituting the air conditioner. inputting data including replacement parts among a plurality of parts constituting the air conditioner as the input data, inferring the characteristics of the worker who repairs the air conditioner ;
The characteristics of the worker include years of experience, skill level related to the outdoor unit of the air conditioner, technical level related to the indoor unit of the air conditioner, technical level related to electronic control, technical level related to machinery, technical level related to communication, and display A diagnosis system for an air conditioner, including a technical level and a technical level for each model of the air conditioner. An air conditioner diagnostic system for remotely diagnosing a plurality of air conditioners,
an inference device for inferring a replacement part among a plurality of parts constituting the air conditioner from input data including an abnormality cause of the air conditioner;
a display device that displays the inferred replacement part,
The reasoning device is
a data acquisition unit that acquires the input data;
inputting the input data acquired by the data acquiring unit into a trained model by deep learning that infers replacement parts among a plurality of parts constituting the air conditioner from the cause of the abnormality of the air conditioner; an inference unit that infers replacement parts among the plurality of parts that make up the harmonic machine,
The inference unit further adds the inferred air conditioner to a trained model by deep learning that infers characteristics of a worker who repairs the air conditioner from replacement parts among a plurality of parts constituting the air conditioner. inputting data including replacement parts among a plurality of parts constituting the air conditioner as the input data, inferring the characteristics of the worker who repairs the air conditioner ;
The characteristics of the worker include years of experience, skill level related to the outdoor unit of the air conditioner, technical level related to the indoor unit of the air conditioner, technical level related to electronic control, technical level related to machinery, technical level related to communication, and display A diagnosis system for an air conditioner, including a technical level and a technical level for each model of the air conditioner. an inference device for inferring characteristics of a worker who repairs the air conditioner from input data including a cause of an abnormality of the air conditioner and a replacement part among a plurality of parts constituting the air conditioner;
a display device that displays the inferred characteristics of the worker;
The reasoning device is
a data acquisition unit that acquires the input data;
The data acquisition unit is added to a trained model by deep learning that infers the characteristics of a worker who repairs the air conditioner from the cause of the abnormality of the air conditioner and the replacement part among the plurality of parts constituting the air conditioner. an inference unit that inputs the input data acquired by and infers data representing the characteristics of the worker who repairs the air conditioner ,
The characteristics of the worker include years of experience, skill level related to the outdoor unit of the air conditioner, technical level related to the indoor unit of the air conditioner, technical level related to electronic control, technical level related to machinery, technical level related to communication, and display A diagnosis system for an air conditioner, including a technical level and a technical level for each model of the air conditioner. An air conditioner diagnostic system for remotely diagnosing a plurality of air conditioners,
an inference device for inferring the cause of an abnormality of the air conditioner from input data including at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner;
a display device that displays the inferred cause of abnormality,
The reasoning device is
a data acquisition unit that acquires the input data;
The data acquisition unit acquires a learned model by deep learning that infers the cause of the abnormality of the air conditioner from at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality of the air conditioner occurs. an inference unit that inputs the input data obtained and infers the cause of the abnormality of the air conditioner,
The inferring unit further adds the inferred cause of the abnormality of the air conditioner to a trained model by deep learning that infers a replacement part among a plurality of parts constituting the air conditioner from the cause of the abnormality of the air conditioner. by inputting the data including
The inference unit further uses deep learning to infer characteristics of a worker who repairs the air conditioner from a cause of abnormality of the air conditioner and a replacement part among a plurality of parts constituting the air conditioner. A worker who repairs the air conditioner by inputting data including the inferred cause of abnormality of the air conditioner and replacement parts among the plurality of parts constituting the air conditioner into the finished model as the input data. infer the
The characteristics of the worker include years of experience, skill level related to the outdoor unit of the air conditioner, technical level related to the indoor unit of the air conditioner, technical level related to electronic control, technical level related to machinery, technical level related to communication, and display A diagnosis system for an air conditioner, including a technical level and a technical level for each model of the air conditioner. According to any one of claims 1 to 9 , comprising a worker selection unit that selects a worker to be dispatched to the repair based on the inferred worker characteristics and the characteristics and schedules of a plurality of workers. A diagnostic system for the described air conditioner. Input data including at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner, and teacher data representing the characteristics of the worker who repairs the air conditioner. a data acquisition unit that acquires learning data that is
Using the learning data, the characteristics of the worker who repairs the air conditioner are inferred from at least the model data of the air conditioner and the operation data of the air conditioner at the time when the abnormality occurred in the air conditioner. a model generation unit that generates a trained model by deep learning for
The characteristics of the worker include years of experience, skill level related to the outdoor unit of the air conditioner, technical level related to the indoor unit of the air conditioner, technical level related to electronic control, technical level related to machinery, technical level related to communication, and display A learning device including a technical level and a technical level for each model of the air conditioner . a data acquisition unit for acquiring learning data composed of input data including an abnormality cause of an air conditioner and teacher data representing characteristics of a worker who repairs the air conditioner;
a model generation unit that generates a trained model by deep learning for inferring the characteristics of a worker who repairs the air conditioner from the cause of an abnormality of the air conditioner using the learning data ,
The characteristics of the worker include years of experience, skill level related to the outdoor unit of the air conditioner, technical level related to the indoor unit of the air conditioner, technical level related to electronic control, technical level related to machinery, technical level related to communication, and display A learning device including a technical level and a technical level for each model of the air conditioner . a data acquisition unit for acquiring learning data composed of input data including replacement parts among a plurality of parts constituting an air conditioner and training data representing characteristics of a worker who repairs the air conditioner;
Using the learning data, a trained model is generated by deep learning for inferring the characteristics of a worker who repairs the air conditioner from replacement parts among a plurality of parts constituting the air conditioner. and a model generator for
The characteristics of the worker include years of experience, skill level related to the outdoor unit of the air conditioner, technical level related to the indoor unit of the air conditioner, technical level related to electronic control, technical level related to machinery, technical level related to communication, and display A learning device including a technical level and a technical level for each model of the air conditioner . Learning data consisting of input data including causes of abnormalities in an air conditioner and replacement parts among a plurality of parts constituting the air conditioner, and training data representing the characteristics of workers who repair the air conditioner. a data acquisition unit that acquires
A deep layer for inferring the characteristics of a worker who repairs the air conditioner from the cause of an abnormality of the air conditioner and replacement parts among a plurality of parts constituting the air conditioner, using the learning data. a model generation unit that generates a trained model by learning ,
The characteristics of the worker include years of experience, skill level related to the outdoor unit of the air conditioner, technical level related to the indoor unit of the air conditioner, technical level related to electronic control, technical level related to machinery, technical level related to communication, and display A learning device including a technical level and a technical level for each model of the air conditioner .

Images