Air Conditioning System and Training Apparatus

Abstract

The air conditioning system comprises: an air conditioner; an inference device to infer data representing a total amount of a power demand value exceeding a set value for a prediction target period from input data including at least one of operation data of the air conditioner for a period prior to the prediction target period, state data of a user of the air conditioner for the period prior to the prediction target period, weather prediction data for the prediction target period, or characteristic data of the room in which the air conditioner is installed; and a control device to cause the air conditioner to perform a heat storage operation depending on a predicted value of the total amount of the power demand value.

Description

TECHNICAL FIELD

The present disclosure relates to an air conditioning system and a training apparatus.

BACKGROUND ART

It has been conventionally known that, in order to prevent a charge for electricity from increasing, a demand controller is used to control a demand for electricity (for example, see PTL 1).

The demand controller has previously registered therein an order in which a plurality of air conditioners having a relatively large load among equipment are stopped when an amount of power exceeds a specified value. When the specified value is exceeded, the demand controller outputs a demand signal to a central processing unit. The central processing unit follows information of the demand signal to transmit a stop signal to an air conditioner via a transmission line in the order in which the plurality of air conditioners are registered until the amount of power is equal to or less than the specified value to thus stop the air conditioners. When a state equal to or less than the specified value continues for a period of time, the central processing unit cancels the stop signal and again operates the air conditioners sequentially.

CITATION LIST

Patent Literature

[PTL 1] Japanese Patent Laying-Open No. 01-118051

SUMMARY OF INVENTION

Technical Problem

While the power demand control described in PTL 1 can prevent an increased charge for electricity, the control stops the air conditioner and thus compromises the comfort of the user.

Therefore, it is an object of the present disclosure to provide an air conditioning system and a training apparatus capable of suppressing power consumption and maintaining comfort without adding a device such as a storage battery.

Solution to Problem

The presently disclosed air conditioning system is an air conditioning system connected to a demand controller to output a demand signal when a power demand value exceeds a set value. The air conditioning system comprises: an air conditioner; an inference device to infer data representing a total amount of the power demand value exceeding the set value for a prediction target period from input data including at least one of operation data of the air conditioner for a period prior to the prediction target period, state data of a user of the air conditioner for the period prior to the prediction target period, weather prediction data for the prediction target period, or characteristic data of a room in which the air conditioner is installed; and a control device to cause the air conditioner to perform a heat storage operation depending on a predicted value of the total amount of the power demand value. In response to the demand signal received from the demand controller, the control device operates the air conditioner using heat stored through the heat storage operation.

The presently disclosed training apparatus is a training apparatus provided for an air conditioning system comprising an air conditioner and communicating with a demand controller. The demand controller outputs a demand signal to the air conditioning system when a power demand value exceeds a set value. The training apparatus comprises: a data obtainer unit to obtain training data including input data including at least one of operation data of the air conditioner for a period prior to a prediction target period, state data of a user of the air conditioner for the period prior to the prediction target period, weather prediction data for the prediction target period, or characteristic data of a room in which the air conditioner is installed, and teacher data including data representing a total amount of the power demand value exceeding the set value for the prediction target period; and a model generation unit using the training data to generate a trained model to infer the data representing the total amount of the power demand value exceeding the set value for the prediction target period from the input data including at least one of the operation data of the air conditioner for the period prior to the prediction target period, the state data of the user of the air conditioner for the period prior to the prediction target period, the weather prediction data for the prediction target period, or the characteristic data of the room in which the air conditioner is installed.

Advantageous Effects of Invention

The presently disclosed air conditioning system infers data representing a total amount of a power demand value exceeding a set value for a prediction target period from input data including at least one of operation data of an air conditioner for a period prior to the prediction target period, state data of a user of the air conditioner for the period prior to the prediction target period, weather prediction data for the prediction target period, or characteristic data of a room in which the air conditioner is installed. The air conditioning system causes the air conditioner to perform a heat storage operation depending on a predicted value of the total amount of the power demand value, and, in response to a demand signal received from a demand controller, operates the air conditioner using heat stored through the heat storage operation. This can suppress power consumption and maintain comfort without adding a device such as a storage battery.

BRIEF DESCRIPTION OF DRAWINGS

FIG. 1 is a diagram showing a system configuration according to a first embodiment.

FIG. 2 is a flowchart showing a procedure of a process performed by an air conditioning system 1 according to the first embodiment.

FIG. 3 is a diagram representing a power demand value and a predicted value P of a total amount of the power demand value that exceeds a set value of a demand controller 3.

FIG. 4 is a table showing an example of input data B1.

FIG. 5 is a table showing an example of teacher data B2.

FIG. 6 is a diagram showing an example of a prediction target period and a period prior to the prediction target period.

FIG. 7 is a table showing an example of operation data of an air conditioner 12.

FIG. 8 is a table showing an example of state data of a user of air conditioner 12.

FIG. 9 is a table showing an example of weather prediction data.

FIG. 10 is a table showing an example of characteristic data of a room in which air conditioner 12 is installed.

FIG. 11 is a diagram showing an exemplary configuration of a neural network.

FIG. 12 is a flowchart for a training process of a training apparatus 2.

FIG. 13 is a table showing an example of prediction data C.

FIG. 14 is a flowchart for an inference process performed by an inference device 16.

FIG. 15 is a diagram showing a hardware configuration of training apparatus 2, inference device 16, or a control device 15.

DESCRIPTION OF EMBODIMENTS

Hereinafter, an embodiment will be described with reference to the drawings.

First Embodiment

FIG. 1 is a diagram showing a system configuration according to a first embodiment.

This system comprises air conditioning systems 1-1 to 1-n, other electric appliances 51-1 to 51-N, a training apparatus 2, a demand controller 3, a wearable terminal 4, and a smartphone 5. In the following description, air conditioning systems 1-1 to 1-n will collectively be referred to as air conditioning system 1, and other electric appliances 51-1 to 51-N will collectively be referred to as another electric appliance 51. Herein, one air conditioning system 1 and another electric appliance 51 constitute a consumer for the sake of illustration.

Demand controller 3 monitors a power demand value of the consumer, and When it is expected that the power demand value will exceed a set value, demand controller 3 outputs a demand signal to air conditioning system 1 to request power suppression for demand control.

Air conditioning system 1 comprises an input device 11, an air conditioner 12, a heat storage unit 13, a communication device 14, a control device 15, an inference device 16, an illuminance sensor 6, an ultraviolet ray sensor 7, and an atmospheric pressure sensor 8.

Input device 11 is for example a remote controller. Input device 11 receives a setting of a target temperature from a user.

Communication device 14 transmits and receives signals to and from demand controller 3, training apparatus 2, wearable terminal 4, and smartphone 5.

Inference device 16 uses a trained model to infer a predicted value of a total amount of the power demand value exceeding a set value of demand controller 3 for a future prediction target period. A prediction target period at a time of inference is a predetermined future period, e.g., the next day, the next week, or the next month.

Training apparatus 2 generates a trained model using past data of the plurality of air conditioning systems 1-1 to 1-n. Training apparatus 2 may be provided on a cloud server.

Heat storage unit 13 includes, for example, a heat storage tank to store a heat storage medium, and a heat exchanger for heat storage.

Air conditioner 12 draws air in a room for which air conditioner 12 is installed to adjust the room's air in temperature and humidity. Air conditioner 12 performs a heat storage operation in accordance with a total amount of the power demand value exceeding the set value of demand controller 3 for a predetermined future period. For example, during the heat storage operation, the heat exchanger for heat storage exchanges heat between the heat storage medium in the heat storage tank and refrigerant passing through a refrigerant circuit of air conditioner 12 to heat or cool the heat storage medium in the heat storage unit.

Control device 15 controls air conditioner 12. Control device 15 causes air conditioner 12 to perform the heat storage operation so that heat storage unit 13 stores heat for a predicted value P of the total amount of the power demand value exceeding the set value of demand controller 3 for the future prediction target period. For example, control device 15 adjusts a time to start the heat storage operation and an operating rate of a compressor based on a heat radiation characteristic of heat storage unit 13 to most efficiently store heat or utilize stored heat. For example, when air conditioner 12 is caused to operate in summer to perform a cooling operation, control device 15 may cause air conditioner 12 to perform the heat storage operation at dawn on the day of the prediction target period so that heat stored in heat storage unit 13 is not radiated. When air conditioner 12 is caused to operate in winter to perform a heating operation, control device 15 may cause air conditioner 12 to perform the heat storage operation during the daytime of the previous day of the prediction target period to allow efficient heat storage.

In response to the demand signal received from demand controller 3, control device 15 operates air conditioner 12 using heat stored in heat storage unit 13 through the heat storage operation.

Illuminance sensor 6 senses the illuminance of the room in which air conditioner 12 is installed. Ultraviolet ray sensor 7 senses an amount of ultraviolet rays in the room in which air conditioner 12 is installed.

Atmospheric pressure sensor 8 senses atmospheric pressure in the room in which air conditioner 12 is installed. FIG. 2 is a flowchart showing a procedure of a process performed by air conditioning system 1 according to the first embodiment.

In step S101, inference device 16 uses a trained model to infer predicted value P of a total amount of the power demand value exceeding the set value of demand controller 3 for a future prediction target period.

In step S102, air conditioner 12 performs the heat storage operation depending on predicted value P of the total amount of the power demand value exceeding the set value of demand controller 3 for the future prediction period.

In step S103, when control device 15 receives the demand signal from demand controller 3, the process proceeds to step S104, whereas when control device 15 does not receive the demand signal from demand controller 3, the process proceeds to step S105.

In step S104, control device 15 operates air conditioner 12 using heat stored in heat storage unit 13 through the heat storage operation.

In step S105, control device 15 causes air conditioner 12 to operate normally. FIG. 3 is a diagram representing a power demand value and predicted value P of a total amount of the power demand value that exceeds the set value of demand controller 3.

For times of 10:00 to 11:00, 11:00 to 12:00, 13:00 to 14:00, 14:00 to 15:00, 15:00 to 16:00, 16:00 to 17:00, and 17:00 to 18:00, the power demand value's predicted value exceeds the set value. Control device 15 generates predicted value P of a total amount of the power demand value for these periods by the heat storage operation in advance. This can prevent power from being used beyond the set value for these periods.

Training apparatus 2 includes a communication device 21, a data obtainer unit 22, a model generation unit 23, and a trained-model storage unit 24.

Data obtainer unit 22 obtains training data including input data B1 and teacher data B2.

FIG. 4 is a table showing an example of input data B1. FIG. 5 is a table showing an example of teacher data B2.

Input data B1 includes at least one of: operation data of air conditioner 12 for a period prior to a prediction target period; state data of a user of air conditioner 12 for the period prior to the prediction target period; weather prediction data for the prediction target period; or characteristic data of the room in which air conditioner 12 is installed.

Teacher data B2 is data representing the total amount of the power demand value exceeding a set value for demand control for the prediction target period.

A prediction target period at a time of training is a fixed past period of time. Teacher data B2 is a total amount of a power demand value actually measured at each point in time in the fixed past period of time. When a future prediction target period at a time of inference is the next day, the next week, and the next month, the past fixed period at the time training is one past day, one past week, and one past month, respectively.

The operation data of air conditioner 12 for a period prior to a prediction target period is used as a factor of the data representing the total amount of the power demand value exceeding the set value for the demand control for the prediction target period because operation data relevant to a load of air conditioner 12 can be assumed to be also the same for the prediction target period.

The state data of the user of air conditioner 12 for a period prior to a prediction target period is used as a factor of the data representing the total amount of the power demand value exceeding the set value for the demand control for the prediction target period because a load by a human body relevant to a load of air conditioner 12 and the user's sensation such as “hot” or “cold” can be assumed to be also the same for the prediction target period.

The weather prediction data for a prediction target period is used as a factor of the data representing the total amount of the power demand value exceeding the set value for the demand control for the prediction target period because the weather of the prediction target period has a large influence on a load of air conditioner 12.

The characteristic data of the room in which air conditioner 12 is installed is used as a factor of the data representing the total amount of the power demand value exceeding the set value for the demand control for the prediction target period because the room's characteristic has a large influence on a load of air conditioner 12.

FIG. 6 is a diagram showing an example of a prediction target period and a period prior to the prediction target period. For example, for a prediction target period X (day D, month M, year Y), a period prior to the prediction target period can be a period A (one day before day D, month M, year Y), a period B (1 week before day D, month M, year Y to one day before day D, month M, year Y), a period C (one month before day D, month M, year Y to one day before day D, month M, year Y), or a period D (one year before day D, month M, year Y to one day before day D, month M, year Y). The prediction target period may be any other period.

FIG. 7 is a table showing an example of the operation data of air conditioner 12. The operation data of air conditioner 12 includes at least one of: external air temperature; indoor set temperature; or an operating rate of a compressor of air conditioner 12.

For example, at least one of: a time-series variation or average value of the external air temperature for period A; a time-series variation or average value of the indoor set temperature for period A; or a time-series variation or average value of the operating rate of the compressor of air conditioner 12 for period A can be used as the operation data. Period A may be replaced with period B, period C, or period D.

FIG. 8 is a table showing an example of the state data of a user of air conditioner 12. The state data of a user of air conditioner 12 includes at least one of: a physical condition of the user in the room in which air conditioner 12 is installed; a state of how the user is present in the room in which air conditioner 12 is installed; an ID of the user present in the room in which air conditioner 12 is installed; or a request by the user for the room in which air conditioner 12 is installed.

The user's physical condition includes, for example, at least one of a pulse rate or a body temperature of the user as sensed via wearable terminal 4.

For example, the user's physical condition can include at least one of an average value of the user's pulse rate for period A or an average value of the user's body temperature for period A. When the user is a plurality of users, a sum or average value of these average values of the plurality of users can be used as the user's physical condition. Period A may be replaced with period B, period C, or period D.

The state of how the user is present in the room includes a period of time for which the user is present in the room. Data obtainer unit 22 can detect whether the user is present in the room based on a schedule of the user registered with a groupwear application, a schedule of an application of smartphone 5 of the user, or data of a GPS of smartphone 5 of the user.

For example, the state of how the user is present in the room may be a period of time in period A for which the user is present in the room. When the user is a plurality of users, a sum or average value of periods of time in period A for which the plurality of users are present in the room can be used as the state of how the user is present in the room. Period A may be replaced with period B, period C, or period D.

The ID of a user present in the room is used to consider that an individual user generates a different thermal load. As well as a period of time for which a user is present in the room, the ID of a user present in the room can be obtained from a schedule of the user registered with a groupwear application, a schedule of an application of smartphone 5 of the user, or data of the GPS of smartphone 5 of the user.

For example, the ID of the user present in the room can be the ID of a user present in the room during period A or the ID of a user present in the room for a fixed period of time or longer during period A. When the user present in the room is a plurality of users, the IDs of the plurality of users can be used as the ID of the user present in the room. Period A may be replaced with period B, period C, or period D.

A request by a user includes, for example, a request sent from the user via smartphone 5 or input device 11 to change the indoor temperature. When the user issues a request once to increase the indoor temperature, the user's request to change it can be +1 (or incremented by one). When the user issues a request once to decrease the indoor temperature, user's request to change it can be −1 (or decremented by one). When the user does not issue a request to change the indoor temperature, the user's request to change it can be 0 (or unchanged).

For example, the request from the user can be a sum of requests sent by the user for period A to change the indoor temperature. When the user is a plurality of users, a sum or average value of requests issued by the plurality of users to change the indoor temperature can be the request by the user. Period A may be replaced with period B, period C, or period D.

FIG. 9 is a table showing an example of weather prediction data. The weather prediction data includes at least one of: a weather forecast (fine, cloudy, rainy) for a local area where air conditioner 12 is installed; an expected temperature of the local area where air conditioner 12 is installed; or a laundry index for the local area where air conditioner 12 is installed.

The laundry index is an index indicating “dryability for laundry.” The laundry index is any one of “dry significantly quick,” “dry quick,” “dry,” “dry less quick,” and “indoor hanging recommended.” “Dry significantly quick” corresponds to a large amount of solar radiation and hence a largest load of air conditioner 12 during the cooling operation. “Indoor drying recommended” corresponds to a small amount of solar radiation and hence a largest load of air conditioner 12 during the heating operation.

For example, when a place where air conditioner 12 is installed is an area A, the weather prediction data includes at least one of a weather forecast, a temperature prediction, or a laundry index for area A for period X.

FIG. 10 is a table showing an example of characteristic data of the room in which air conditioner 12 is installed. The characteristic data of the room in which air conditioner 12 is installed includes at least one of: a type of ventilation in the room in which air conditioner 12 is installed; the atmospheric pressure in the room in which air conditioner 12 is installed; the amount of ultraviolet rays in the room in which air conditioner 12 is installed; or the illuminance of the room in which air conditioner 12 is installed.

The type of ventilation can be expressed by, for example, a ratio of ventilation by a lossless outdoor-air treating air conditioner relative to energy recovery ventilation (ventilation by the lossless outdoor-air treating air conditioner and ventilation by equipment other than the lossless outdoor-air treating air conditioner) (hereinafter referred to as an “outdoor-air treating, air conditioning ratio”). When the lossless outdoor-air treating air conditioner is not installed, the outdoor-air treating, air conditioning ratio is 0. As the outdoor-air treating, air conditioning ratio increases, a load of air conditioner 12 and a power demand value of a consumer decrease.

Atmospheric pressure sensor 8 can sense atmospheric pressure. Atmospheric pressure is used because as atmospheric pressure is low, air is rarefied, and air conditioner 12 will be less efficient. Thus, air conditioner 12 in a high-rise building and an area of a high altitude is less efficient. As atmospheric pressure decreases, a load of air conditioner 12 and a power demand value of a consumer increase, and the load of air conditioner 12 peaks earlier.

Illuminance sensor 6 can sense illuminance. For the cooling operation, when illuminance is large, a load of air conditioner 12 and a power demand value of a consumer increase, and the load of air conditioner 12 peaks earlier. For the heating operation, when illuminance is large, the load of air conditioner 12 and the power demand value of the consumer decrease.

Ultraviolet ray sensor 7 can sense an amount of ultraviolet rays. For the cooling operation, when the amount of ultraviolet rays is a large amount, a load of air conditioner 12 and a power demand value of a consumer increase, and the load of air conditioner 12 peaks earlier. For the heating operation, when the amount of ultraviolet rays is a large amount, the load of air conditioner 12 and the power demand value of the consumer decrease.

Referring to FIG. 1 again, based on training data including input data B1 and teacher data B2 output from data obtainer unit 22, model generation unit 23 generates a trained model by learning B2 input when B1 is input.

The training data is data in which input data B1 and teacher data B2 are associated with each other.

Model generation unit 23 can use a learning algorithm which is a supervised learning algorithm. As an example, an example in which a neural network is applied will be described.

Model generation unit 23 generates a trained model by so-called supervised learning for example in accordance with a neural network model. Herein, supervised learning refers to a method in which a set of input data B1 and teacher data B2 is provided to a training apparatus so that B2 input when B1 is input is learnt.

The neural network includes an input layer including a plurality of neurons, an intermediate layer (or a hidden layer) including a plurality of neurons, and an output layer including a plurality of neurons. The intermediate layer may be one layer or two or more layers.

FIG. 11 is a diagram showing an exemplary configuration of the neural network. For example, for a three-layer neural network as shown in FIG. 11, when a plurality of inputs are input to the input layer (X1-X3), the inputs have their values multiplied by a weight W1 (w11-w16) and thus input to the intermediate layer (Y1-Y2), and a result thereof is further multiplied by a weight W2 (w21-w26) and thus output from the output layer (Z1-Z3). The output result changes depending on the values of weights W1 and W2.

In the present embodiment, the neural network is trained as W1 and W2 are adjusted so that when input data B1 is input to the input layer and the output layer outputs a result closer to teacher data B2.

Model generation unit 23 generates a trained model through such a training as described above and outputs the trained model to trained-model storage unit 24. Specifically, as the trained model, weights W1 and W2 are output to trained-model storage unit 24.

The learning algorithm used by model generation unit 23 can use deep learning for learning extraction of an exact feature value and may perform machine learning according to other known methods, e.g., genetic programming, functional logic programming, support vector machine, or the like.

Trained-model storage unit 24 stores the trained model output from model generation unit 23.

Hereinafter, a training process performed by training apparatus 2 will be described. FIG. 12 is a flowchart for a training process of training apparatus 2.

In step S201, data obtainer unit 22 obtains input data B1 and teacher data B2. While data obtainer unit 22 simultaneously obtains input data B1 and teacher data B2 for the sake of illustration, receiving input data B1 and teacher data B2 in association with each other suffices, and data obtainer unit 22 may obtain input data B1 and teacher data B2, as differently timed.

In step S202, in accordance with training data created based on a combination of input data B1 and teacher data B2 obtained by data obtainer unit 22, model generation unit 23 trains a model by so-called supervised learning to output B2 when B1 is received to thus generate a trained model.

In step S203, trained-model storage unit 24 stores the trained model generated by model generation unit 23.

Referring to FIG. 1 again, inference device 16 includes a data obtainer unit 17, an inference unit 18, and a trained-model storage unit 19.

The trained model stored in trained-model storage unit 24 of training apparatus 2 is transferred to trained-model storage unit 19.

Data obtainer unit 17 obtains new input data B1 which is not used for training. Inference unit 18 infers prediction data C obtained using the trained model.

FIG. 13 is a table showing an example of prediction data C. Prediction data C is data representing a predicted value of a total amount of a power demand value exceeding the set value for the demand control for a prediction target period.

Inference unit 18 inputs new input data B1 that is obtained at data obtainer unit 17 to the trained model to output prediction data C inferred from new input data B1.

Hereinafter, a process for inferring prediction data C using inference device 16 will be described. FIG. 14 is a flowchart for an inference process performed by inference device 16. This process more specifically describes step S101 in FIG. 2.

In step S301, data obtainer unit 17 obtains new input data B1. In step S302, inference unit 18 inputs input data B1 to the trained model stored in trained-model storage unit 19, and obtains prediction data C.

In step S303, inference unit 18 outputs to control device 15 prediction data C obtained through the trained model.

Exemplary Variation

(1) Inference device 16, training apparatus 2, and control device 15 described in the first embodiment can have an equivalent operation configured by hardware of digital circuitry or software. When a function of inference device 16, training apparatus 2, and control device 15 is implemented by using software, inference device 16, training apparatus 2, and control device 15 can for example include a processor 5002 and a memory 5001 connected by a bus 5003 as shown in FIG. 15 and a program stored in memory 5001 can be executed by processor 5002.

(2) Data obtainer unit 22 of training apparatus 2 may obtain training data from a plurality of air conditioning systems used in the same area, or may obtain training data from a plurality of air conditioning systems operating independently in different areas.

Data obtainer unit 22 of training apparatus 2 can add to or remove from a target an air conditioning system for which training data is collected.

(3) Training apparatus 2 may use data of an air conditioning system A to generate a trained model, and the trained model may be used by an inference device of that air conditioning system A. In this case, input data B1 may not include characteristic data of the room in which air conditioner 12 is installed.

(4) Training apparatus 2 may use data of an air conditioning system A to generate a trained model, and the trained model may be used as an initial value by an inference device of another air conditioning system B. Air conditioning system B uses data of air conditioning system B to retrain the trained model.

(5) Inference device 16 may reside on a cloud server. (6) Teacher data B2 and prediction data C may include other data in addition to a total amount of a power demand value exceeding the set value for the demand control for a prediction target period.

For example, teacher data B2 and prediction data C may include a load of air conditioner 12. The load of air conditioner 12 and the total amount of the power demand value exceeding the set value for the demand control change inherently in the same manner. When how the total amount of the power demand value exceeding the set value for the demand control that is output from inference device 16 changes is significantly different from how the load of air conditioner 12 changes, it can be determined that inappropriate training has been performed.

Alternatively, teacher data B2 and prediction data C may include data representing a time at which the power demand value exceeds the set value for the demand control. Based on the time at which the power demand value exceeds the set value for the demand control, control device 15 may determine a time at which air conditioner 12 starts the heat storage operation.

Teacher data B2 and prediction data C may not exactly be a total amount of a power demand value exceeding the set value for the demand control for a prediction target period, and may be data representing the same.

For example, teacher data B2 and prediction data C may be data of a power demand value for each time in the prediction target period. As the set value is a fixed value, even such data can also be said to represent the total amount of the power demand value exceeding the set value for the demand control.

(7) Input data B1 is not limited to that described in the first embodiment. For example, the operation data in input data B1 may include data which distinguishes the cooling operation and the heating operation from each other, data representing the rotational speed of the compressor, data representing performance in operation of the air conditioner, or data representing power consumption of the air conditioner.

(8) While in the above-described embodiment the inference device uses a trained model to output data representing a predicted value of a total amount of a power demand value exceeding a set value for a prediction target period from input data obtained by a data obtainer unit, this is not exclusive.

For example, the inference device may be based on rule-based reasoning or case-based reasoning to output data representing a predicted value of a total amount of a power demand value exceeding a set value for a prediction target period from input data obtained by the data obtainer unit.

It should be understood that the embodiments disclosed herein have been described for the purpose of illustration only and in a non-restrictive manner in any respect. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the meaning and scope equivalent to the terms of the claims.

REFERENCE SIGNS LIST

1 air conditioning system, 2 training apparatus, 3 demand controller, 4 wearable terminal, 5 smartphone, 6 illuminance sensor, 7 ultraviolet ray sensor, 8 air pressure sensor, 11 input device, 12 air conditioner, 13 heat storage unit, 14, 21 communication unit, 15 control device, 16 inference device, 17, 22 data obtainer unit, 18 inference unit, 19, 24 trained-model storage unit, 23 model generation unit, 51 other electric appliance, 5001 memory, 5002 processor, 5003 bus.

Claims

1. An air conditioning system connected to a demand controller to output a demand signal when a power demand value exceeds a set value, the air conditioning system comprising:

an air conditioner; an inferencer to infer data representing a total amount of the power demand value exceeding the set value for a prediction target period from input data including at least one of operation data of the air conditioner for a period prior to the prediction target period, state data of a user of the air conditioner for the period prior to the prediction target period, weather prediction data for the prediction target period, or characteristic data of a room in which the air conditioner is installed; and

a controller to cause the air conditioner to perform a heat storage operation depending on a predicted value of the total amount of the power demand value, wherein

in response to the demand signal received from the demand controller, the controller operates the air conditioner using heat stored through the heat storage operation.

2. The air conditioning system according to claim 1, wherein the inferencer includes:

a data obtainer to obtain the input data including at least one of the operation data of the air conditioner for the period prior to the prediction target period, the state data of the user of the air conditioner for the period prior to the prediction target period, the weather prediction data for the prediction target period, or the characteristic data of the room in which the air conditioner is installed; and

an inferencer, using a trained model to infer the data representing the total amount of the power demand value exceeding the set value for the prediction target period from the input data including at least one of the operation data of the air conditioner for the period prior to the prediction target period, the state data of the user of the air conditioner for the period prior to the prediction target period, the weather prediction data for the prediction target period, or the characteristic data of the room in which the air conditioner is installed, to output from the input data obtained by the data obtainer data representing the predicted value of the total amount of the power demand value exceeding the set value for the prediction target period.

3. The air conditioning system according to claim 2, wherein the operation data of the air conditioner includes at least one of an external air temperature, an indoor set temperature, or an operating rate of a compressor of the air conditioner.

4. The air conditioning system according to claim 2, wherein the state data of the user of the air conditioner includes at least one of: a physical condition of the user in the room in which the air conditioner is installed; a state of how the user is present in the room in which the air conditioner is installed; an ID of the user present in the room in which the air conditioner is installed; or a request by the user for the room in which the air conditioner is installed.

5. The air conditioning system according to claim 4, wherein the physical condition of the user includes at least one of a pulse rate or a body temperature of the user as detected by a wearable terminal.

6. The air conditioning system according to claim 4, wherein the state of how the user is present in the room includes a period of time for which the user is present in the room.

7. The air conditioning system according to claim 4, wherein the data obtainer detects whether the user is present in the room based on a schedule of the user registered with a groupwear application, a schedule of an application of a smartphone of the user, or data of a GPS of the smartphone of the user.

8. The air conditioning system according to claim 2, wherein the weather prediction data includes at least one of: a weather forecast for a local area where the air conditioner is installed; an expected temperature of the local area where the air conditioner is installed; or a laundry index for the local area where the air conditioner is installed.

9. The air conditioning system according to claim 2, wherein the characteristic data of the room in which the air conditioner is installed includes at least one of: a type of ventilation in the room in which the air conditioner is installed; atmospheric pressure in the room in which the air conditioner is installed; an amount of ultraviolet rays in the room in which the air conditioner is installed; or illuminance of the room in which the air conditioner is installed.

10. The air conditioning system according to claim 9, wherein the type of ventilation is represented by a ratio of ventilation by a lossless outdoor-air treating air conditioner to energy recovery ventilation.

11. A training apparatus provided for an air conditioning system comprising an air conditioner and communicating with a demand controller, wherein the demand controller outputs a demand signal to the air conditioning system when a power demand value exceeds a set value, the training apparatus comprising:

a data obtainer to obtain training data including input data including at least one of operation data of the air conditioner for a period prior to a prediction target period, state data of a user of the air conditioner for the period prior to the prediction target period, weather prediction data for the prediction target period, or characteristic data of a room in which the air conditioner is installed, and teacher data including data representing a total amount of the power demand value exceeding the set value for the prediction target period; and

a model generator using the training data to generate a trained model to infer the data representing the total amount of the power demand value exceeding the set value for the prediction target period from the input data including at least one of the operation data of the air conditioner for the period prior to the prediction target period, the state data of the user of the air conditioner for the period prior to the prediction target period, the weather prediction data for the prediction target period, or the characteristic data of the room in which the air conditioner is installed.

12. The training apparatus according to claim 11, wherein the operation data of the air conditioner includes at least one of an external air temperature, an indoor set temperature, or an operating rate of a compressor of the air conditioner.

13. The training apparatus according to claim 11, wherein the state data of the user of the air conditioner includes at least one of: a physical condition of the user in the room in which the air conditioner is installed; a state of how the user is present in the room in which the air conditioner is installed; an ID of the user present in the room in which the air conditioner is installed; or a request by the user for the room in which the air conditioner is installed.

14. The training apparatus according to claim 13, wherein the physical condition of the user includes at least one of a pulse rate or a body temperature of the user as detected by a wearable terminal.

15. The training apparatus according to claim 13, wherein the state of how the user is present in the room includes a period of time for which the user is present in the room.

16. The training apparatus according to claim 13, wherein the data obtainer detects whether the user is present in the room based on a schedule of the user registered with a groupwear application, a schedule of an application of a smartphone of the user, or data of a GPS of the smartphone of the user.

17. The training apparatus according to claim 11, wherein the weather prediction data includes at least one of: a weather forecast for a local area where the air conditioner is installed; an expected temperature of the local area where the air conditioner is installed; or a laundry index for the local area where the air conditioner is installed.

18. The training apparatus according to claim 11, wherein the characteristic data of the room in which the air conditioner is installed includes at least one of: a type of ventilation in the room in which the air conditioner is installed; atmospheric pressure in the room in which the air conditioner is installed; an amount of ultraviolet rays in the room in which the air conditioner is installed; or illuminance of the room in which the air conditioner is installed.

19. The training apparatus according to claim 18, wherein the type of ventilation is represented by a ratio of ventilation by a lossless outdoor-air treating air conditioner to energy recovery ventilation.