Information processing system, method, and program

Abstract

To improve accuracy of predicting an operating state from operational data deviated from a predetermined definition area.SOLUTION: According to one embodiment of this disclosure, an information processing system is an information processing system including a prediction model for predicting an operating state of an apparatus from operational data of the apparatus, and control part. The control part detects that operation data during operation inputted to the prediction mode is deviated from a predetermined definition area, determines an operating state of the apparatus when the operational data during operation is returned to the definition area after the detection, and re-learns the prediction model with the deviated operational data and a result of determination of an operating state of the apparatus during deviation from the definition area created on the basis of the result of the determination when the definition area is returned as teacher data for re-learning.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to an information processing system, a method, and a program. [Background technology]

[0002] Conventionally, it is known to use a predictive model generated by machine learning to predict the operating state of equipment from operating data during operation of the equipment, thereby controlling the operation of the equipment or diagnosing faults in the equipment (Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6791429 Summary of the Invention [Problem to be solved by the invention]

[0004] However, if the operating data during operation is outside a predetermined range (eg, the range of the operating data used to generate the prediction model), it is difficult to accurately predict the operating state from the operating data.

[0005] An object of the present disclosure is to improve the accuracy of predicting a driving state from driving data that deviates from a predetermined domain. [Means for solving the problem]

[0006] An information processing system according to a first aspect of the present disclosure includes: An information processing system including a prediction model for predicting an operating state of an equipment from operating data of the equipment, and a control unit, The control unit is Detecting that operational data during operation input to the prediction model deviates from a predetermined domain; When the operational data during operation returns to the domain after the detection, the operational state of the device is determined; The predictive model is re-learned using the deviating operating data and the judgment result of the operating state of the equipment when it deviates from the domain, which is created based on the judgment result when the equipment returns, as training data for re-learning.

[0007] According to the first aspect of the present disclosure, it is possible to improve the accuracy of predicting a driving state from driving data that deviates from a predetermined domain.

[0008] An information processing system according to a second aspect of the present disclosure is the information processing system according to the first aspect, The control unit is Supplying past operation data of the equipment and an index value for determining the operation state of the equipment calculated from the operation data as teacher data; generating the prediction model that outputs the index value when the driving data of the teacher data is input; Acquire operational data of the device during operation; a prediction value of the index value obtained by inputting the operational data during the operation into the prediction model and an actual measurement value of the index value calculated from the operational data during the operation are compared to determine an operational state of the equipment, and an operation command is sent to the equipment based on the result of the determination; When it is detected that the operational data during operation deviates from a predetermined domain, the deviating operational data is stored in a storage unit; inputting the operation data that has returned to the domain after the detection of the deviating operation data into the prediction model to determine whether the operation state of the equipment is normal or abnormal; When the operating state of the equipment is normal, the deviating operating data is determined to be normal, and the prediction model is re-learned using the deviating operating data as teacher data for re-learning.

[0009] According to the second aspect of the present disclosure, a predictive model can be generated using only normal driving data.

[0010] An information processing system according to a third aspect of the present disclosure is the information processing system according to the first or second aspect, The range that deviates from the domain is outside the distribution range of the training data that has already been trained for the prediction model.

[0011] According to the third aspect of the present disclosure, it is possible to improve the accuracy of predicting a driving state from driving data outside the distribution range of learned learning data of a prediction model.

[0012] An information processing system according to a fourth aspect of the present disclosure is the information processing system according to the first or second aspect, The range deviating from the domain is a partial range within the distribution range of the training data for which the prediction model has been trained.

[0013] According to the fourth aspect of the present disclosure, it is possible to improve the accuracy of predicting a driving state from driving data within a partial range within a distribution range of learned learning data of a prediction model.

[0014] An information processing system according to a fifth aspect of the present disclosure is the information processing system according to any one of the first to fourth aspects, The control unit is The predictive model is re-learned using the deviated operating data acquired by another device installed at a different location from the device on which the predictive model was generated.

[0015] According to the fifth aspect of the present disclosure, it is possible to generate a prediction model suitable for another device installed in a different location from the device for which the prediction model was generated.

[0016] An information processing system according to a sixth aspect of the present disclosure is the information processing system according to any one of the first to fourth aspects, Multiple devices share the same prediction model, The control unit is Re-learning is performed using the deviating operating data acquired by each of the plurality of devices.

[0017] According to the sixth aspect of the present disclosure, a prediction model suitable for each of a plurality of devices can be generated.

[0018] An information processing system according to a seventh aspect of the present disclosure is the information processing system according to any one of the first to sixth aspects, The equipment is a refrigeration and air conditioning equipment.

[0019] According to the seventh aspect of the present disclosure, it is possible to improve the accuracy of predicting the operating state of a refrigeration and air conditioning equipment from operating data of the refrigeration and air conditioning equipment that deviates from a predetermined domain.

[0020] An information processing system according to an eighth aspect of the present disclosure is the information processing system according to the seventh aspect, The operating data is at least one of the following: a condenser outlet subcooling degree, a compressor suction temperature, a compressor discharge temperature, an outside air temperature, a room temperature, a refrigeration or air conditioning capacity, and a compressor rotation speed.

[0021] According to an eighth aspect of the present disclosure, it is possible to improve the accuracy of predicting the operating state when at least one of the condenser outlet subcooling degree, the compressor suction temperature, the compressor discharge temperature, the outside air temperature, the room temperature, the refrigeration or air conditioning capacity, and the compressor rotation speed deviates from a predetermined defined domain.

[0022] An information processing system according to a ninth aspect of the present disclosure is the information processing system according to the seventh aspect, The operating condition is a leakage of refrigerant from the refrigeration and air conditioning equipment.

[0023] According to the ninth aspect of the present disclosure, it is possible to improve the accuracy of predicting refrigerant leakage from refrigeration and air conditioning equipment based on operation data of the refrigeration and air conditioning equipment that deviates from a predetermined domain.

[0024] A method according to a tenth aspect of the present disclosure, comprising: A method executed by a control unit of an information processing system including a prediction model for predicting an operating state of an equipment from operating data of the equipment, the method comprising: Detecting that operational data during operation input to the prediction model deviates from a predetermined domain; When the operational data during operation returns to the domain after the detection, the operational state of the device is determined; The predictive model is re-learned using the deviating operating data and the judgment result of the operating state of the equipment when it deviates from the domain, which is created based on the judgment result when the equipment returns, as training data for re-learning.

[0025] A program according to an eleventh aspect of the present disclosure, A prediction model for predicting an operating state of an apparatus from operating data of the apparatus, and a control unit of an information processing system including the control unit, A step of detecting whether operational data during operation input to the prediction model deviates from a predetermined domain; a step of determining an operating state of the device when the operational data during operation returns to the domain after the detection; The apparatus executes a procedure for re-learning the prediction model using the deviated operating data and the judgment result of the operating state of the equipment when it deviates from the domain, which is created based on the judgment result when the equipment returns, as training data for re-learning. [Brief description of the drawings]

[0026] [Figure 1] FIG. 1 is a diagram for explaining an outline of the present disclosure. [Diagram 2] 1 illustrates an example of the overall configuration of the present disclosure. [Diagram 3] FIG. 13 is a diagram for explaining the outline of the present disclosure using another embodiment. [Figure 4] FIG. 1 is a hardware configuration diagram of an information processing system according to an embodiment of the present disclosure. [Diagram 5]FIG. 1 is a hardware configuration diagram of an air conditioning system (in cooling operation) according to an embodiment of the present disclosure. [Figure 6] FIG. 2 is a hardware configuration diagram of an air conditioning system (in heating operation) according to an embodiment of the present disclosure. [Figure 7] FIG. 1 is a hardware configuration diagram of an air conditioning system (in the case of simultaneous cooling and heating operation) according to an embodiment of the present disclosure. [Figure 8] 13 is a flowchart of a re-learning process according to an embodiment of the present disclosure. [Figure 9] FIG. 13 is a diagram for explaining a range deviating from a predetermined domain according to an embodiment of the present disclosure. [Figure 10] FIG. 13 is a diagram for explaining a range deviating from a predetermined domain according to an embodiment of the present disclosure. [Figure 11] FIG. 1 is a diagram for explaining an apparatus for generating a predictive model and an apparatus for relearning a predictive model according to an embodiment of the present disclosure. [Figure 12] FIG. 13 is a diagram for explaining sharing of a prediction model by a plurality of devices according to an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0027] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings.

[0028] <Terminology explanation> "Operation data" refers to any data that can be acquired during the operation of a device. For example, the operation data of a device (e.g., a refrigeration and air conditioning device) is as follows: - Information related to the state of the refrigerant circulating inside the refrigeration and air conditioning equipment, such as temperature (condenser outlet temperature, compressor suction temperature, compressor discharge temperature, etc.), pressure (condensation pressure, evaporation pressure), and circulation volume. Values ​​calculated from refrigerant temperature and pressure data (condenser outlet supercooling, compressor suction superheat, etc.). The operation amount and control command value of the actuator that controls the state of the refrigerant (expansion valve opening, compressor rotation speed, fan rotation speed, compressor suction superheat target value, etc.). -Time information regarding the operation status of equipment (compressor operation time, defrost operation time, etc.). -Environmental conditions where the equipment is operating (outdoor temperature, room temperature, indoor humidity, etc.). Input energy to the equipment, such as compressor current, fan current, etc. The refrigeration or air conditioning capacity that is the output of the equipment. The operating data used in this disclosure is at least one of the condenser outlet subcooling degree, compressor suction temperature, compressor discharge temperature, outside air temperature, room temperature, refrigeration or air conditioning capacity, and compressor rotation speed. An "operating state" is a state of an equipment (e.g., an index value for determining the operating state of the equipment) used to control the operation of the equipment or diagnose equipment failures. For example, the operating state is the amount of refrigerant held. If the amount of refrigerant held is a specified value, the equipment is in a normal state, and if the amount of refrigerant held is equal to or less than the specified value, the equipment is in an abnormal state such as a gas shortage or leakage.

[0029] <Summary> 1 is a diagram for explaining an outline of the present disclosure, which will be explained with reference to an information processing system 20 and a device 30 in FIG.

[0030] <<Generating a predictive model>> An information processing system 20 (details will be described later) generates a prediction model that predicts an operating state (for example, a normal operating state, an abnormal operating state, or whether the operating state is normal or abnormal) from operating data of a device 30 such as a refrigeration and air conditioning device (details will be described later). Specifically, the information processing system 20 generates a prediction model by machine learning using a set of past normal operating data x0 of the device 30, which is an air conditioner shown in 11 of FIG. 1, as initial learning data. Then, as shown in 12 of FIG. 1, the generated prediction model is used to determine the refrigerant reserve of the air conditioner. The initial learning data 11 is a plurality of operation data sets acquired between past times T=ts and te. In this example, the data set at each time includes four items of data (rcp: compressor rotation speed, Ta: outside air temperature, Tc: condensation temperature, Tb: condenser outlet temperature). In this example, SC, defined as the difference between the condensation temperature Tc and the condenser outlet temperature Tb, is calculated as an evaluation index of the refrigerant reserve. In model learning, a prediction model is generated by learning the correlation between the compressor speed rcp obtained directly from x0, the outside air temperature Ta, and the SC calculated from the data of x0. When the compressor speed rcp and the outside air temperature Ta are input, the generated prediction model outputs SC', which is the predicted value of SC at that time.

[0031] <<Determining refrigerant reserves using a prediction model>> The information processing system 20 judges the amount of refrigerant held according to the flow shown in 13 of FIG. 1. First, the operating data x of the device 30 during operation is acquired. x has the same data structure as the initial learning data x0, and includes rcp: compressor rotation speed, Ta: outside air temperature, Tc: condensation temperature, and Tb: condenser outlet temperature. The device 30 inputs rcp and Ta into the generated prediction model and predicts SC', which is the predicted value of SC under normal conditions. The device 30 also calculates SC as the difference between Tc and Tb. Furthermore, it calculates ΔSC, which is the difference between SC and SC'. ΔSC is an index that corrects the change in SC due to the compressor rotation speed and the outside air temperature and represents only the change in the amount of refrigerant held. Therefore, if the amount of refrigerant held by the device 30 is a specified amount, ΔSC is almost 0, and if the amount of refrigerant held decreases below the specified amount, ΔSC decreases and becomes a negative value. When ΔSC is input in the judgment, the refrigerant held amount judgment result y is output. y is the result of judging the amount of refrigerant stored, and for example, if the amount is normal, it is labeled as "normal" or "0". For example, if the amount is less than the specified amount, it is labeled as "leak" or "1".

[0032] <<Detection of driving data that deviates from a predefined domain>> During operation of the device 30, the information processing system 20 judges whether the operation data of the device 30 during operation deviates (x_out) or not (x_in) from a predetermined domain. For example, 13-1 in FIG. 1 shows a schematic representation of the relationship between x_in and x_out when Ta in the x item is 35° C. or higher and falls outside the learning range. When the information processing system 20 detects that the operation data of the device 30 during operation deviates from a predetermined domain, it stores the operation data (x_out).

[0033] <<Detection of driving data returning to a predefined domain>> Thereafter, when the information processing system 20 detects that the operation data during the operation of the device 30 has returned to a predetermined domain, it judges the operation state by inputting the first returned operation data (x_inr) into the prediction model. 13-1 in FIG. 1 shows the relationship between x_out and x_inr. For example, the information processing system 20 stores a judgment result y_inr calculated from a predicted value of the operation state output by inputting x_inr into the prediction model. If the judgment y_inr for the first operation data (x_inr) that has returned to the predetermined domain is normal, the information processing system 20 judges that the judgment result y_out for the input x_out at the time of deviation is also normal. Even if the judgment result of y_out is abnormal, if y_inr is normal, the judgment result is corrected to normal. The relationship between y_in, y_out, and y_inr is also shown in 13-1 in FIG. 1. In this example, x_out is data at one point in time, but it may be multiple data that are continuous in time.

[0034] <<Relearning>> When the determination result y_inr of the driving state when the driving data returns to a predetermined domain is normal, the information processing system 20 determines that the stored x_out is normal data. Then, as shown in 14 of FIG. 1, re-learning data is created by adding x_out to the initial learning data x0. Thereafter, re-learning is performed using the created re-learning data to update the prediction model. Note that re-learning may be performed each time x_out is added, or may be performed after waiting for multiple x_out to be added.

[0035] The prediction model in the above embodiment is a model (model predicting a normal driving state) generated using normal driving data, and only normal driving data is used during re-learning. In addition, the prediction model may be a model (model predicting an abnormal driving state) generated using abnormal driving data, or a model (model predicting whether the driving state is normal or abnormal) generated using normal driving data and abnormal driving data (driving states labeled as normal or abnormal). In addition, only abnormal driving data may be used during re-learning, or normal driving data and abnormal driving data (driving states labeled as normal or abnormal) may be used. As another embodiment, a case in which normal driving data and abnormal driving data are used will be described with reference to FIG. 3. In this example, the initial learning data (11 in FIG. 3) is a data item x0 in the example in FIG. 1, to which y, which is a judgment label for normal and abnormal, is added. In the model learning flow (12 in FIG. 3), four items of driving data (rcp, Ta, Tc, Tb) and judgment labels are used as learning data in model learning. The generated prediction model inputs four items of driving data and outputs a judgment result y. Since this model is a model for predicting normality or abnormality, the judgment result y includes a label of normality or abnormality. In the flow of actual operation (13 in FIG. 3), the operating data during actual operation is the same as in FIG. 1, and is four items: rcp, Ta, Tc, and Tb. The four items of data are directly input to the prediction model, and the output of the prediction model is the judgment result y, so that the state quantity calculation (SC calculation), state quantity deviation calculation (ΔSC calculation), and result judgment are omitted. When data x_out that deviates from a predetermined domain is input, the information processing system 20 detects the deviation and stores x_out and its judgment result y_out. After that, when the information processing system 20 detects that the operating data during operation of the device 30 has returned to the predetermined domain, it stores the judgment result y_inr for the first returned operating data x_inr. Then, the value of y_out is changed to the value of y_inr. In the creation of driving data for re-learning (14 in FIG. 3), the driving data for re-learning is created by adding the deviated driving data x_out and the judgment result y_out at the time of deviation to the initial learning data x0. After that, re-learning is performed using this data.

[0036] <Overall configuration example> 2 shows an example of the overall configuration of the present disclosure. An information processing system 20 and a device 30 are communicably connected via an arbitrary network. The information processing system 20 includes a prediction model that predicts the operating state of the device 30 from the operating data of the device 30, and a control unit (control unit 1 in FIG. 4).

[0037] The information processing system 20 can determine the operating state of the device 30 by inputting the operating data of the device 30 into a prediction model during operation of the device 30. For example, the information processing system 20 determines the operating state of the device 30 by comparing a predicted value of the operating state (e.g., a predicted value of a normal operating state or a predicted value of an abnormal operating state) output when the operating data of the device 30 is input into the prediction model with an actual measured value of the operating state calculated from the operating data of the device 30. The information processing system 20 transmits an operation command to the device 30 based on the result of the determination of the operating state of the device 30.

[0038] The information processing system 20 re-learns using operational data during operation that deviates from a predetermined domain, and updates the prediction model.

[0039] The equipment 30 is any equipment such as a refrigeration / air conditioning equipment.

[0040] In addition, the information processing system 20 may be implemented on a cloud server separate from the device 30, or may be implemented on a computer installed in the same building or the like as the device 30, or may be implemented as part of the device 30.

[0041] <Hardware configuration of information processing system> FIG. 4 is a hardware configuration diagram of the information processing system 20 according to an embodiment of the present disclosure.

[0042] The information processing system 20 has a control unit (e.g., a CPU (Central Processing Unit)) 1, a ROM (Read Only Memory) 2, and a RAM (Random Access Memory) 3. The control unit 1, the ROM 2, and the RAM 3 form a so-called computer. The information processing system 20 may also have an auxiliary storage device 4, a display device 5, an operation device 6, and an I / F (Interface) device 7. The hardware components of the information processing system 20 are connected to each other via a bus 8.

[0043] The control unit 1 is a computing device that executes various programs installed in the auxiliary storage device 4 .

[0044] The ROM 2 is a non-volatile memory. The ROM 2 functions as a main storage device that stores various programs, data, etc. required for the control unit 1 to execute various programs installed in the auxiliary storage device 4. Specifically, the ROM 2 functions as a main storage device that stores boot programs such as a BIOS (Basic Input / Output System) and an EFI (Extensible Firmware Interface).

[0045] The RAM 3 is a volatile memory such as a dynamic random access memory (DRAM) or a static random access memory (SRAM). The RAM 3 functions as a main storage device that provides a working area in which various programs installed in the auxiliary storage device 4 are expanded when the control unit 1 executes them.

[0046] The auxiliary storage device 4 is an auxiliary storage device that stores various programs and information used when the various programs are executed.

[0047] The display device 5 is a display device that displays the internal state of the information processing system 20 and the like.

[0048] The operation device 6 is an input device through which an administrator of the information processing system 20 inputs various instructions to the information processing system 20 .

[0049] The I / F unit 7 is a communication device that connects to various sensors and networks and communicates with other terminals.

[0050] Hereinafter, the hardware configuration of an air conditioning system 100, which is an example of the device 30, will be described with reference to Figs. 5 to 7. The air conditioning system 100 may be any air conditioning system, such as a multi-air conditioner for a building, a central air conditioning system using a chiller as a heat source, an air conditioner for a store or office, or a room air conditioner, and may be a refrigeration or freezing system other than for cooling and heating purposes. The air conditioning system 100 may have a plurality of indoor units 300. The plurality of indoor units 300 may include indoor units with different performance, may include indoor units with the same performance, or may include indoor units that are stopped.

[0051] <Hardware configuration of air conditioning system (for cooling operation)> 5 is a hardware configuration diagram of an air conditioning system (in cooling operation) 100 according to an embodiment of the present disclosure. The air conditioning system 100 has an outdoor unit 200 and one or more indoor units 300.

[0052] In the example of FIG. 5, the outdoor heat exchanger 201, the outdoor unit main expansion valve 205, the subcooling heat exchanger 203, the indoor heat exchanger expansion valve 302, the four-way switching valve 206, the indoor heat exchanger 301, and the compressor 202 are connected by refrigerant piping to form a main refrigerant circuit. The four-way switching valve 206 is set to supply the discharge gas of the compressor 202 to the outdoor heat exchanger 201. In the example of FIG. 5, a subcooling heat exchanger expansion valve 204 is further provided in a bypass piping connected from the piping between the outdoor heat exchanger 201 and the subcooling heat exchanger 203 to the piping on the suction side of the compressor 202. The subcooling heat exchanger 203 is a heat exchanger that exchanges heat between the refrigerant that has passed through the subcooling heat exchanger expansion valve 204 provided in the bypass piping connected from the piping between the outdoor heat exchanger 201 and the subcooling heat exchanger 203 to the piping on the suction side of the compressor 202 and the refrigerant in the main refrigerant circuit. The bypass example in FIG. 5 is just one example.

[0053] <<Outdoor unit>> On the outdoor unit 200 side, an outdoor heat exchanger 201, a compressor 202, a subcooling heat exchanger 203, a subcooling heat exchanger expansion valve (bypass circuit) 204, and an outdoor unit main expansion valve (main refrigerant circuit) 205 are connected to piping. The outdoor unit 200 has various sensors (temperature sensors (e.g., thermistors) (1), (3), (4), (6), (7) and pressure sensors (2), (5), etc.).

[0054] <<Indoor unit>> On the indoor unit 300 side, an indoor heat exchanger 301 and an indoor heat exchanger expansion valve 302 are connected to piping. The indoor unit 300 has various sensors (temperature sensors (e.g., thermistors) (8), (9), etc.).

[0055] <Hardware configuration of air conditioning system (for heating operation)> 6 is a hardware configuration diagram of an air conditioning system (in heating operation) 100 according to an embodiment of the present disclosure. The air conditioning system 100 has an outdoor unit 200 and one or more indoor units 300.

[0056] 6, the outdoor heat exchanger 201, the compressor 202, the four-way switching valve 206, the indoor heat exchanger 301, the indoor heat exchanger expansion valve 302, the subcooling heat exchanger 203, and the outdoor unit main expansion valve 205 are connected by refrigerant piping to configure a main refrigerant circuit. The four-way switching valve 206 has a flow path set so as to supply the discharge gas of the compressor 202 to the indoor heat exchanger 301.

[0057] <<Outdoor unit>> On the outdoor unit 200 side, an outdoor heat exchanger 201, a compressor 202, a subcooling heat exchanger 203, a subcooling heat exchanger expansion valve (bypass circuit) 204, and an outdoor unit main expansion valve (main refrigerant circuit) 205 are connected to piping. The outdoor unit 200 has various sensors (temperature sensors (e.g., thermistors) (1), (3), (4), (6), (7) and pressure sensors (2), (5), etc.).

[0058] <<Indoor unit>> On the indoor unit 300 side, an indoor heat exchanger 301 and an indoor heat exchanger expansion valve 302 are connected to piping. The indoor unit 300 has various sensors (temperature sensors (e.g., thermistors) (8), (9), etc.).

[0059] <Hardware configuration of air conditioning system (for simultaneous cooling and heating operation)> The present disclosure is not limited to application to cooling operation and heating operation, but can also be applied to simultaneous cooling and heating operation. Simultaneous cooling and heating operation will be described below with reference to FIG.

[0060] Fig. 7 is a hardware configuration diagram of an air conditioning system (in the case of simultaneous cooling and heating operation) 100 according to an embodiment of the present disclosure. In the air conditioning system 100, outdoor heat exchangers 201-1 and 201-2, which are divided into two parts, and multiple indoor units are connected by three communication pipes, and simultaneous cooling and heating operation is possible. Fig. 7 shows an example of cooling-dominated operation, in which the indoor unit 300-1 is operated in heating mode and the indoor unit 300-2 is operated in cooling mode. At this time, the outdoor heat exchanger 201-1 functions as a condenser, and the outdoor heat exchanger 201-2 functions as an evaporator.

[0061] For example, the equipment 30 is a refrigeration and air conditioning equipment. The air conditioning equipment described above as an air conditioning system is used for indoor cooling and heating. Equipment is selected according to the purpose from cooling only types and heating only types without a four-way switching valve, cooling and heating switching types with a four-way switching valve, and simultaneous cooling and heating operation types. Refrigeration equipment includes products such as refrigerators and ice makers used in the refrigeration and freezing fields, supermarket showcases, refrigerated warehouses, and refrigerated containers. Refrigeration and air conditioning equipment is basically equipment that produces and uses cold energy, so there is no four-way switching valve. Some refrigeration and air conditioning equipment has an economizer circuit or an intermediate injection circuit. Economizers and intermediate injection are omitted in the basic circuit as a way to improve performance.

[0062] <Method> 8 is a flowchart of a re-learning process according to an embodiment of the present disclosure. The control unit 1 of the information processing system 20 executes the following re-learning process.

[0063] The information processing system 20 includes a prediction model that predicts the operating state of the device 30 (for example, a normal operating state, an abnormal operating state, or whether the operating state is normal or abnormal) from the operating data of the device 30. The prediction model is generated using past operating data of the device 30 (input data is "past operating data of the device 30", and output data is "an operating state calculated from the past operating data of the device 30").

[0064] In step 101 (S101), the control unit 1 of the information processing system 20 acquires operation data during operation of the device 30 (e.g., refrigeration and air conditioning device). For example, the operation data is at least one of the following: condenser outlet subcooling degree, compressor suction temperature, compressor discharge temperature, outside air temperature, room temperature, refrigeration or air conditioning capacity, and compressor rotation speed.

[0065] The control unit 1 of the information processing system 20 can determine the operating state (e.g., refrigerant leakage) by continuously acquiring operating data during operation of the device 30 and inputting the operating data into a prediction model. For example, the control unit 1 of the information processing system 20 compares a predicted value of the operating state (e.g., a predicted value of a normal operating state or a predicted value of an abnormal operating state) output when the operating data of the device 30 is input into the prediction model with an actual measured value of the operating state calculated from the operating data of the device 30 to determine the operating state of the device 30. The control unit 1 of the information processing system 20 transmits an operation command to the device 30 based on the result of the determination of the operating state of the device 30.

[0066] In step 102 (S102), the control unit 1 of the information processing system 20 detects that the driving data acquired in S101 deviates from a predetermined domain.

[0067] In step 103 (S103), the control unit 1 of the information processing system 20 stores the driving data detected in S102 that deviates from the predetermined domain in a storage unit (for example, a memory of the information processing system 20).

[0068] In step 104 (S104), the control unit 1 of the information processing system 20 detects, after the detection in S103, that the driving data (the driving data that the control unit 1 of the information processing system 20 continues to acquire) has returned to a predetermined domain.

[0069] In step 105 (S105), the control unit 1 of the information processing system 20 judges the driving state by inputting the driving data that has returned to the predetermined domain in S104 into a prediction model.

[0070] For example, the control unit 1 of the information processing system 20 inputs the driving data that has returned to a predetermined domain into a prediction model and outputs a predicted value of the driving state. The control unit 1 of the information processing system 20 also calculates the driving state from the driving data that has returned to the predetermined domain. The information processing system 20 compares the predicted value of the driving state with the actual measured value of the driving state to determine whether the driving state when the driving data has returned to the predetermined domain is normal or abnormal.

[0071] In step 106 (S106), the control unit 1 of the information processing system 20 performs re-learning using the driving data that deviated from the predetermined domain, and updates the prediction model.

[0072] Specifically, the control unit 1 of the information processing system 20 determines that the driving state when the driving data returns to the predetermined domain is normal, and that the driving data that deviated from the predetermined domain was normal. On the other hand, the control unit 1 of the information processing system 20 determines that the driving state when the driving data returns to the predetermined domain is abnormal, and that the driving data that deviated from the predetermined domain was abnormal. Then, the information processing system 20 re-learns using the driving data that deviated from the predetermined domain and the driving state calculated from the driving data (data used to generate the prediction model may also be used) and updates the prediction model.

[0073] When relearning, only normal driving data may be used, only abnormal driving data may be used, or both normal and abnormal driving data (driving states labeled as normal or abnormal) may be used.

[0074] <Example of operation data> Here, a description will be given of an example of the operating data of the device 30. For example, the operating data is at least one of the condenser outlet subcooling degree, the compressor suction temperature, the compressor discharge temperature, the outside air temperature, the room temperature, the refrigeration or air conditioning capacity, and the compressor rotation speed.

[0075] (Example 1) For example, the operation data of the equipment is Condensation temperature Evaporation temperature Condenser outlet temperature Evaporator outlet temperature Outside temperature - Compressor 202 rotation speed Opening degree of the expansion valve 204 of the subcooling heat exchanger Current value of compressor 202 It may include at least one of the following:

[0076] (Example 2) For example, the operation data of the equipment may be, in addition to or instead of the above operation data (Example 1), Opening degree of indoor unit expansion valve 302 Opening degree of outdoor unit main expansion valve 205 - Total rated capacity of indoor units in operation or standby Number of indoor units in operation ·Indoor capacity (cooling or heating) Indoor unit outlet temperature ·room temperature - Outdoor unit liquid shutoff valve connecting piping refrigerant temperature (connecting piping liquid temperature detected by thermistor (4) in Figures 5 and 7) Liquid connection pipe refrigerant temperature (temperature measured at the connection pipe outside the outdoor unit 200 detected by an external sensor attached to the outside of the outdoor unit 200) Outdoor unit fan air volume Indoor unit fan air volume Outdoor unit fan speed (step, tap) Indoor unit fan speed (step, tap) Outdoor unit fan current value Indoor unit fan current value ·Refrigerant circulation amount Discharge temperature of compressor 202 Compressor 202 suction temperature Compressor 202 discharge superheat Compressor 202 suction superheat Degree of subcooling at the outlet of the subcooling heat exchanger 203 (if a subcooling heat exchanger circuit is provided) Outlet superheat of the subcooling heat exchanger 203 (gas pipe side) (if a subcooling heat exchanger circuit is included) Economizer outlet subcooling degree (if economizer circuit is installed) Economizer expansion valve opening (if an economizer circuit is included) Economizer bypass outlet pressure (if economizer circuit is installed) - Opening degree of intermediate injection expansion valve (if intermediate injection circuit is provided) - Intermediate injection temperature (if intermediate injection circuit is included) - Intermediate injection pressure (if intermediate injection circuit is installed) Evaporator inlet water temperature (when at least one of the heat source side and the user side is water-cooled) - Evaporator outlet water temperature (when at least one of the heat source side and the user side is water-cooled) -Condenser inlet water temperature (when at least one of the heat source side and the user side is water-cooled) - Condenser outlet water temperature (when at least one of the heat source side and the user side is water-cooled) It may include at least one of the following:

[0077] (Example 3) For example, the operating data for inferring the predicted value of the refrigerant amount index value that is strongly correlated with the amount of refrigerant held under normal conditions may be, in addition to or instead of the operating data described above (Examples 1 and 2), the following: At least one of the number of defrosts and the defrost time may be included.

[0078] <Example of operating state> Here, an example of the operating state of the device will be described.

[0079] (Example 1 (cooling operation)) For example, the operating state is Condensation temperature-Outlet temperature of the outdoor heat exchanger 201 (hereinafter also referred to as the outdoor heat exchanger outlet subcooling degree. Note that the subcooling degree is also called SC or subcool) Compressor intake superheat (superheat is also called SH or superheat) Compressor discharge superheat - Value based on the outdoor heat exchanger outlet subcooling degree or compressor suction superheat degree or compressor discharge superheat degree It may include at least one of the following:

[0080] For example, the value based on the outdoor heat exchanger outlet subcooling degree is a calculated value using the outdoor heat exchanger outlet subcooling degree. For example, Calculated value using the degree of subcooling at the outlet of the outdoor heat exchanger = degree of subcooling at the outlet of the outdoor heat exchanger / (condensing temperature - outdoor temperature) It is.

[0081] For example, the value based on the degree of subcooling at the outdoor heat exchanger outlet is a value defined from the refrigerant property values ​​or their calculated values ​​of the part related to the degree of subcooling at the outdoor heat exchanger outlet on the refrigeration cycle diagram (TS, Ph diagram).

[0082] (Example 2 (cooling operation)) For example, the operating state may be, in addition to the above refrigerant amount index value (Example 1), or instead of the outdoor heat exchanger outlet subcooling degree of the above refrigerant amount index value (Example 1), ·Subcooling heat exchanger outlet subcooling degree - Value based on the degree of subcooling at the outlet of the subcooling heat exchanger It may include at least one of the following:

[0083] (Example 3 (Heating operation)) In the case of heating operation, the operation state is changed from the above operation state (example 1 and example 2) to the following: The degree of subcooling at the indoor heat exchanger outlet may include at least one of the following: a degree of subcooling at the indoor heat exchanger outlet, and a value based on the degree of subcooling at the indoor heat exchanger outlet. The degree of subcooling at the indoor heat exchanger outlet is at least one of the degrees of subcooling at the multiple indoor heat exchangers 301, an average of the degrees of subcooling at the multiple indoor heat exchangers 301, or a degree of subcooling at the indoor side junction or the outdoor side junction of the multiple indoor heat exchangers 301.

[0084] (Example 4 (simultaneous cooling and heating operation)) In the case of simultaneous cooling and heating operation, in addition to the above operating states (at least one of Example 1 and Example 2), A combination of the outlet subcooling degree of an indoor heat exchanger (the indoor heat exchanger 301 of the heating indoor unit 300-1 in FIG. 7) and the outlet subcooling degree of an outdoor heat exchanger (the outdoor heat exchanger (condenser) 201-1 in FIG. 7).

[0085] Hereinafter, examples of ranges deviating from the predetermined domain will be described with reference to FIGS.

[0086] <Example 1> 9 is a diagram for explaining a range deviating from a predetermined domain according to an embodiment of the present disclosure. In an embodiment of the present disclosure, the range deviating from a predetermined domain is outside the distribution range of the learning data that has been learned by the prediction model. Hereinafter, a description will be given with reference to FIG.

[0087] 9 shows the distribution range of operating data during operation and the distribution range of learning data (operating data used to generate a prediction model) when the operating data is parameter 1 (e.g., outside air temperature) and parameter 2 (e.g., compressor rotation speed). In Example 1, when the operating data during operation is outside the distribution range of the learning data, it can be used as operating data for re-learning.

[0088] <Example 2> 10 is a diagram for explaining a range deviating from a predetermined domain according to an embodiment of the present disclosure. In an embodiment of the present disclosure, the range deviating from a predetermined domain is a part of a distribution range of the learning data of the prediction model. Hereinafter, a description will be given with reference to FIG. 10.

[0089] 10 shows the distribution range of the operating data during operation and the distribution range of the learning data (operating data used to generate a prediction model) when the operating data is parameter 1 (e.g., outside air temperature) and parameter 2 (e.g., compressor rotation speed). In Example 2, even if the operating data during operation is within a partial range of the distribution range of the learning data, it can be used as operating data for re-learning if there is a reason, for example, to increase the amount of learning data in that operating range because the amount of data is originally insufficient.

[0090] <<Devices for generating predictive models and devices for retraining predictive models>> Fig. 11 is a diagram for explaining an apparatus for generating a prediction model and an apparatus for re-learning a prediction model according to an embodiment of the present disclosure. In an embodiment of the present disclosure, the control unit 1 of the information processing system 20 can re-learn using deviating driving data acquired by another apparatus (e.g., an apparatus of the same model as the apparatus for generating the prediction model) installed in a different location from the apparatus for generating the prediction model. Hereinafter, a description will be given with reference to Fig. 11.

[0091] It is assumed that the control unit 1 of the information processing system 20 generates a prediction model using driving data acquired from a device (device A in FIG. 11) installed at a certain location (location A in FIG. 11). After that, the control unit 1 of the information processing system 20 can re-learn using driving data that deviates from a predetermined domain and that is acquired from another device (device B in FIG. 11) installed at a location (location B in FIG. 11) different from the location (location A in FIG. 11) of the device (device A in FIG. 11) used to generate the prediction model.

[0092] << Sharing of prediction models by multiple devices >> Fig. 12 is a diagram for explaining sharing of a prediction model by multiple devices according to an embodiment of the present disclosure. In an embodiment of the present disclosure, multiple devices (e.g., multiple devices of the same model) share the same prediction model, and the control unit 1 of the information processing system 20 can re-learn using deviating operating data acquired by each of the multiple devices. Hereinafter, a description will be given with reference to Fig. 12.

[0093] Assume that multiple devices (devices B1, B2, and B3 in FIG. 12) share the same prediction model. The control unit 1 of the information processing system 20 can re-learn using operation data that deviates from a predetermined domain acquired by each of the multiple devices (devices B1, B2, and B3 in FIG. 12). Note that in one embodiment of the present disclosure, the multiple devices (devices B1, B2, and B3 in FIG. 12) are installed in a location different from the location of the devices used to generate the prediction model.

[0094] Although the embodiments have been described above, it will be understood that various changes in form and details are possible without departing from the spirit and scope of the claims. [Explanation of symbols]

[0095] 1. Control section 2 ROM 3 RAM 4 Auxiliary storage 5 Display device 6 Operating device 7 I / F device 8 Bus 11 Initial learning data 12 Flowchart of predictive model generation 13 Actual operation flow chart 13-1 Relationship between data within the domain and data outside the domain 14 Retraining data 20 Information Processing Systems 30 equipment 100 Air Conditioning System 200 outdoor unit 201 Outdoor heat exchanger 201-1 Outdoor heat exchanger (condenser) 201-2 Outdoor heat exchanger (evaporator) 202 Compressor 203 Supercooling heat exchanger 204 Subcooling heat exchanger expansion valve 205 Outdoor unit main expansion valve 206 Four-way switching valve 300 indoor unit 300-1 Indoor heating unit 300-2 Cooling indoor unit 301 Indoor heat exchanger 302 Indoor heat exchanger expansion valve

Claims

1. An information processing system including a prediction model for predicting an operating state of an equipment from operating data of the equipment, and a control unit, The control unit is Detecting that operational data during operation input to the prediction model deviates from a predetermined domain; When the operational data during operation returns to the domain after the detection, the operational state of the device is determined; An information processing system that re-learns the prediction model using as training data for re-learning the deviated operating data acquired by another device installed at a different location from the device for which the prediction model was generated, and the result of a judgment of the operating state of the device when it deviates from the domain of definition, which is created based on the result of the judgment at the time of recovery.

2. The control unit is Supplying past operation data of the equipment and an index value for determining the operation state of the equipment calculated from the operation data as teacher data; generating the prediction model that outputs the index value when the driving data of the teacher data is input; Acquire operational data of the device during operation; a prediction value of the index value obtained by inputting the operational data during the operation into the prediction model and an actual measurement value of the index value calculated from the operational data during the operation are compared to determine an operational state of the equipment, and an operation command is sent to the equipment based on the result of the determination; When it is detected that the operational data during operation deviates from a predetermined domain, the deviating operational data is stored in a storage unit; inputting the operation data that has returned to the domain after the detection of the deviating operation data into the prediction model to determine whether the operation state of the equipment is normal or abnormal; When the operating state of the device is normal, the deviating operating data is determined to be normal, and the prediction model is retrained using the deviating operating data as training data for retraining. The information processing system according to claim 1 .

3. The information processing system according to claim 1 or 2, wherein the range deviating from the domain is outside a distribution range of training data for which the prediction model has already been trained.

4. The information processing system according to claim 1 , wherein the range deviating from the domain is a part of a distribution range of training data for which the prediction model has already been trained.

5. Multiple devices share the same prediction model, The control unit is The information processing system according to claim 1 , further comprising: re-learning using the deviating driving data acquired by each of the plurality of devices.

6. The information processing system according to claim 1 , wherein the equipment is a refrigeration and air conditioning equipment.

7. The information processing system according to claim 6 , wherein the operating data is at least one of a condenser outlet subcooling degree, a compressor suction temperature, a compressor discharge temperature, an outside air temperature, a room temperature, a refrigeration or air conditioning capacity, and a compressor rotation speed.

8. The information processing system according to claim 6 , wherein the operating condition is a leakage of refrigerant from the refrigeration and air conditioning equipment.

9. A method executed by a control unit of an information processing system including a prediction model for predicting an operating state of an equipment from operating data of the equipment, the method comprising: Detecting that operational data during operation input to the prediction model deviates from a predetermined domain; When the operational data during operation returns to the domain after the detection, the operational state of the device is determined; A method for re-learning the prediction model using as training data for re-learning the deviated operating data acquired by another device installed at a different location from the device for which the prediction model was generated, and the result of a judgment of the operating state of the device when it deviates from the domain of definition, which is created based on the result of the judgment at the time of recovery.

10. A prediction model for predicting an operating state of an apparatus from operating data of the apparatus, and a control unit of an information processing system including the control unit, A step of detecting whether operational data during operation input to the prediction model deviates from a predetermined domain; a step of determining an operating state of the device when the operational data during operation returns to the domain after the detection; A program for executing a procedure for re-learning the prediction model using, as training data for re-learning, the deviating operating data acquired by another device installed at a different location from the device for which the prediction model was generated, and the result of a judgment of the operating state of the device when it deviates from the domain, which is created based on the result of the judgment at the time of recovery.

Images