Information processing apparatus, charging system, power control method, and program

Abstract

To determine the probability distribution of predicted values of power consumption, provide information on the accuracy of the predicted values, and control a load system.SOLUTION: An information processing apparatus comprises: an acquisition unit that acquires relevant information that is information predetermined as information affecting the power consumption of a load system; a prediction unit that inputs the relevant information in a prediction execution period for performing prediction, to a machine learning model generated by performing machine learning with the relevant information in the past power consumption period as an explanatory variable, and the power consumption of the load system in the past power consumption period as an objective variable, and thereby predicts the probability distribution of the power consumption of the load system in a prediction target period during which a result of prediction is applied; and a control unit that, on the basis of the probability distribution of the power consumption of the load system in the prediction target period, restricts the power consumption of the load system in the prediction target period.SELECTED DRAWING: Figure 4

Description

Technical Field

[0001] The present invention relates to an information processing apparatus, a charging system, a power control method, and a program.

Background Art

[0002] In Patent Document 1, for the purpose of improving the accuracy of predicting the power consumption of target customers, a computer is caused to execute a step of acquiring first calendar information indicating a prediction target period, and using, as explanatory variables, second calendar information indicating the date and time when power is consumed by a first power load whose operation is controlled by a control device or a second power load different from the first power load in at least one customer, and performing machine learning with the power consumption obtained by subtracting the power consumption of the first power load from the sum of the power consumption of the first power load and the power consumption of the second power load at the date and time as the objective variable, and inputting the first calendar information into the generated machine learning model to predict the power consumption of the target customer in the prediction target period. A program for causing the computer to execute the step is disclosed.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] When controlling the power consumption of a customer based on a prediction using machine learning or the like, if the accuracy of the prediction is low, there is a high possibility that control will be implemented to limit the power consumption even though the actual power consumption is less than the threshold, or control will be implemented not to limit the power consumption even though the actual power consumption is greater than the threshold.

[0005] The object of the present invention is to obtain the probability distribution of the predicted value of power consumption, provide information on the accuracy of the predicted value, and control load equipment.

Means for Solving the Problems

[0006] The invention according to claim 1 includes an acquisition unit that acquires related information, which is information determined in advance as information that affects the power consumption of load equipment, and uses the related information at the time of past power consumption as an explanatory variable, and the power consumption of the load equipment at the time of past power consumption as an objective variable. A prediction unit that predicts the probability distribution of the power consumption of the load equipment at the time of prediction target to which the result of the prediction is applied by inputting the related information at the time of prediction execution for machine learning, and based on the probability distribution of the power consumption of the load equipment at the time of the prediction target, It is an information processing apparatus characterized by including a control unit that restricts the power consumption of the load equipment at the time of the prediction target. The invention according to claim 2 is the information processing apparatus according to claim 1, characterized in that the related information includes at least any one of season, date, day of the week, time, weather information, event information of the region to which the load equipment belongs, usage status of the facility to which the load equipment belongs, and usage status of the load equipment. The invention according to claim 3 is the information processing apparatus according to claim 2, characterized in that the acquisition unit acquires the related information at the time of prediction execution. The invention according to claim 4 includes an acquisition means that acquires related information, which is information determined in advance as information that affects the power consumption of load equipment including charging equipment, and uses the related information at the time of past power consumption as an explanatory variable, and the power consumption of the load equipment at the time of past power consumption as an objective variable. A prediction means that predicts the probability distribution of the power consumption of the load equipment at the time of prediction target to which the result of the prediction is applied by inputting the related information at the time of prediction execution for machine learning, and based on the probability distribution of the power consumption of the load equipment at the time of the prediction target, It is a charging system characterized by including a control means for determining the charging power of the charging equipment. The invention according to claim 5 is the charging system according to claim 4, characterized in that the control means operates the charging equipment determined in advance based on the probability distribution of the power consumption of the load equipment at the time of prediction target. The invention according to claim 6 is the charging system according to claim 4, characterized in that the control means does not limit the charging equipment when the remaining battery level of the electric vehicle charged by the charging equipment is less than a predetermined charging amount. The invention according to claim 7 is a power control method having: a step of acquiring related information which is information determined in advance as information affecting the power consumption of the load equipment; a step of inputting the related information at the time of prediction execution for prediction into a machine learning model generated by performing machine learning using the related information at the time of past power consumption as an explanatory variable and the power consumption of the load equipment at the time of past power consumption as an objective variable, thereby predicting the probability distribution of the power consumption of the load equipment at the time of prediction target to which the result of the prediction is applied; and a step of restricting the power consumption of the load equipment at the time of prediction target based on the probability distribution of the power consumption of the load equipment at the time of prediction target. The invention according to claim 8 is a program for causing a computer to realize: a function of acquiring related information which is information determined in advance as information affecting the power consumption of the load equipment; a function of predicting the probability distribution of the power consumption of the load equipment at the time of prediction target to which the result of the prediction is applied by inputting the related information at the time of prediction execution for prediction into a machine learning model generated by performing machine learning using the related information at the time of past power consumption as an explanatory variable and the power consumption of the load equipment at the time of past power consumption as an objective variable; and a function of restricting the power consumption of the load equipment at the time of prediction target based on the probability distribution of the power consumption of the load equipment at the time of prediction target.

Advantages of the Invention

[0007] According to the present invention, it is possible to obtain the probability distribution of the predicted value of the power consumption, provide information on the accuracy of the predicted value, and control the load equipment.

Brief Description of the Drawings

[0008]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Mode for Carrying Out the Invention

[0009] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. First, the electricity charge will be described. The electricity charge is mainly composed of a power consumption charge and a basic charge. Among them, the basic charge is calculated based on the basic charge unit price and the contract power. The contract power is the maximum value of the maximum demand power within one year retroactive from the current month. The maximum demand power is the maximum value of the average power consumption for each monthly time limit (demand time limit: 30 minutes). The average power consumption is the average value of the power consumption at each time limit.

[0010] As described above, the contract power is the maximum value of the maximum demand power within the past year. Therefore, when the maximum demand power of a certain month becomes the contract power, even if the maximum demand power lower than this contract power is continuously maintained after that month, the basic charge based on this contract power will be imposed for one year. Also, when the average power consumption at a certain time period exceeds the value of the contract power up to that point and becomes the maximum demand power of the month including this time period, the average power consumption (maximum demand power) at this time period will be used as the new contract power for calculating the basic charge thereafter.

[0011] Therefore, the power consumption at a certain time period is predicted, and based on the prediction result, control is performed to reduce the power consumption so that the average power consumption at that time period does not exceed the value of the contract power. Hereinafter, the time when the prediction is executed is referred to as the prediction execution time, and the time when the prediction result is applied is referred to as the prediction target time.

[0012] <Configuration of the System> FIG. 1 is a diagram showing the configuration of a system 1 to which the present embodiment is applied. The system 1 shown in FIG. 1 is composed of, for example, an information processing device 2, a control device 4 installed in a facility 3 of a single consumer (hereinafter, the facility of the consumer is referred to as the "consumer facility"), and load equipment 5. In this specification, a "consumer" refers to a person who uses electric power. Also, a "consumer facility" refers to a facility in which devices, equipment, etc. that use electric power are installed. Also, "load equipment" refers to something that includes one or more devices, equipment, etc. that use electric power. Electric power is supplied from the power grid 9 shown in FIG. 1 to the load equipment 5 in the consumer facility 3.

[0013] The information processing device 2 is a management server that manages the power consumption of the load facility 5. The information processing device 2 predicts the probability distribution of the power consumption amount of the load facility 5 at the time to be predicted, and restricts the power consumption of the load facility 5 based on the predicted probability distribution of the power consumption amount of the load facility 5. A method for restricting the power consumption of the load facility 5 will be described later. The information processing device 2 may be realized by a virtual server provided in a cloud computing environment. Further, it may be realized by a single or a plurality of server machines connectable to a network.

[0014] As described above, the contract power is based on the average power consumption every demand time limit (30 minutes). Therefore, in the present embodiment, the time unit to be predicted is set to 30 minutes in accordance with the above demand time limit. Note that this time unit is merely an example, and other times (in minutes, hours, days, etc.) may be used as the time unit.

[0015] The customer facility 3 is a facility in which devices, equipment, etc. that use power are installed as described above. Examples of the customer facility 3 include, for example, houses, offices, factories, and the like. The load facility 5 represents one or a plurality of devices and equipment that use power as described above. Examples of the individual devices and equipment included in the load facility 5 include a load device 6, a storage battery 7, a charging facility 8, and the like. The control device 4 is installed in the customer facility 3 to receive an instruction from the information processing device 2 and control the power consumption amount of the individual devices and equipment included in the load facility 5.

[0016] The load device 6 refers to a device that uses power. As an example of the load device 6, for example, when the customer facility 3 is a house, an air conditioner, lighting, a personal computer, etc. can be mentioned. Further, for example, when the customer facility 3 is a factory, in addition to the above loads, production equipment in a manufacturing factory, etc. can also be considered as an example of the load device 6. Further, the storage battery 7 refers to a device that stores power and discharges the power when necessary. The storage battery 7 is used, for example, as a power source or an emergency power source that replaces the power grid 9. The charging facility 8 refers to a device that supplies power to a storage battery to be charged for charging. Examples of the object to be charged include, for example, electric vehicles.

[0017] The information processing device 2 and the control device 4 are connected via a network. Here, the network is a communication means used for information communication between the information processing device 2 and the control device 4. For example, a LAN (Local Area Network) is used. Also, it may be connected using the Internet or a public line.

[0018] <Hardware Configuration of Information Processing Device> FIG. 2 is a diagram showing a hardware configuration example of the information processing device 2. The information processing device 2 is provided with an arithmetic processing unit 51, a storage device 52 for storing various kinds of information, and a communication interface 53 for communicating with an external device. The arithmetic processing unit 51 is constituted by a computer. The arithmetic processing unit 51 has a CPU (Central Processing Unit) 51a as an example of a processor that executes various processes described later. Also, the arithmetic processing unit 51 has a ROM (Read Only Memory) 51b in which a program is stored and a RAM (Random Access Memory) 51c used as a work area. The storage device 52 is realized by an existing device such as a hard disk drive or a semiconductor memory. The arithmetic processing unit 51 and the storage device 52 are connected through a bus 54 and signal lines (not shown). Also, the information processing device 2 in the present embodiment may have a display device 55 and an input device 56. As the display device 55, for example, a display or the like is used. As the input device 56, for example, a keyboard, a mouse, a touch panel, or the like is used.

[0019] The program executed by the CPU 51a can be provided to the information processing device 2 in a state of being stored in a computer-readable recording medium such as a magnetic recording medium (magnetic tape, magnetic disk, etc.), an optical recording medium (optical disk, etc.), a magneto-optical recording medium, or a semiconductor memory. Also, the program executed by the CPU 51a may be provided to the information processing device 2 using a communication means. The program provided to the information processing device 2 is stored in the storage device 52.

[0020] In this specification, the processor refers to a processor in a broad sense, including general-purpose processors (e.g., CPU: Central Processing Unit, etc.) and dedicated processors (e.g., GPU: Graphics Processing Unit, ASIC: Application Specific Integrated Circuit, FPGA: Field Programmable Gate Array, programmable logic devices, etc.). Further, the operation of the processor may be performed not only by one processor but also by a plurality of physically separated processors cooperating with each other. Also, the order of each operation of the processor is not limited to the order described in this embodiment and may be changed.

[0021] <Functional Configuration of Information Processing Apparatus> FIG. 3 is a diagram showing a functional configuration example of the information processing apparatus 2. The information processing apparatus 2 in this embodiment includes a communication unit 21, an acquisition unit 22, a generation unit 23, a storage unit 24, a prediction unit 25, a determination unit 26, and a control unit 27. Further, the information processing apparatus 2 may include a UI (User Interface) unit 28. The UI unit 28 notifies the user of information and accepts input of information from the user. Each of these functions is realized by the cooperation of hardware and software. For example, the function of the communication unit 21 is realized by the communication interface 53. Also, the operations by the acquisition unit 22, generation unit 23, prediction unit 25, determination unit 26, control unit 27, etc. are performed by the CPU 51a executing a program stored in the storage device 52. Further, the storage unit 24 stores data and programs necessary for the CPU 51a to perform processing in the RAM 51c and the storage device 52. The UI unit 28 is composed of a display device 55 and an input device 56. Hereinafter, each function will be described.

[0022] <Function of Communication Unit> The communication unit 21 communicates with an external device, for example, the control device 4 in the customer facility 3, and transmits and receives various types of information and instructions.

[0023] <Function of the acquisition unit> The acquisition unit 22 acquires information used for generating a machine learning model described later and information used for prediction using the machine learning model. The information used for generating the machine learning model and the information used for prediction are information predetermined as information that affects the power consumption of the load facility 5. Further, the acquisition unit 22 acquires the power usage history and log information of the customer as information used for generating the machine learning model. The acquisition unit 22 is an example of an acquisition means.

[0024] Examples of the information predetermined as information that affects the power consumption of the load facility 5 include, for example, season, date, day of the week, time, weather information, event information of the region to which the load facility 5 belongs, the usage status of the facility to which the load facility 5 belongs, and the usage status of the load facility 5. Note that depending on the type of the load facility 5, the information considered to affect its power consumption may be different. In this case, the acquisition unit 22 may acquire only the information necessary for prediction according to the type of the load facility 5 that is the target of predicting the power consumption. Hereinafter, the information predetermined as information that affects the power consumption of the load facility 5 may be referred to as related information.

[0025] The power consumption of the load facility 5 varies, for example, by season. More specifically, for example, when the season is summer, the temperature becomes higher compared to the case of spring. In such a case, the user demands a comfortable environment, the demand for air conditioners and the like increases, and the power consumption increases. Therefore, the information on the season can be used as related information.

[0026] Also, the power consumption of the load facility 5 varies depending on the date and day of the week. For example, on weekdays, when users etc. work at the business office, the power consumption of that business office increases, but on holidays, since the users are not at the business office, the power consumption of that business office becomes smaller compared to weekdays. Therefore, the information on the date and day of the week can be used as related information.

[0027] In addition, the power consumption of the load equipment 5 also varies depending on the time period. For example, during the daytime to evening, when users are active, the power consumption tends to increase. On the other hand, at night, since user activities decrease, the power consumption tends to decrease. Specifically, during the time period from late at night to early morning, user activities decrease, so the power consumption decreases. Therefore, the time period information can be used as related information.

[0028] In addition, the power consumption of the load equipment 5 also varies depending on the weather information. The weather information includes weather, temperature, humidity, etc. For example, when the weather is snowy or the temperature is low, users seek a comfortable environment, and the demand for air conditioners, heaters, etc. increases, resulting in an increase in power consumption. Also, when the humidity is high, users seek a comfortable environment, and the use of dehumidifying air conditioners, etc. causes an increase in power consumption. Therefore, the weather information can be used as related information. The weather information can be obtained via a network from a weather information providing means, such as the Meteorological Agency.

[0029] In addition, the power consumption of the load equipment 5 also varies depending on the presence or absence of events, etc. When an event is held in the area to which the load equipment 5 belongs, during the event period, the power consumption of the load equipment 5 increases. More specifically, for example, when an event such as a live performance is held, many users visit the area to participate in the event. Therefore, for example, the need to operate an air conditioning device arises. Also, since audio devices, video devices, etc. operate for a long time, during the event period, the power consumption of the load equipment 5 increases. Therefore, the information on the presence or absence of events can be used as related information. The event information in the area to which the load equipment 5 belongs can be obtained in advance via a network, etc.

[0030] Also, the power consumption of the load facility 5 varies depending on the usage status of the customer facility 3 to which the load facility 5 belongs. Here, when explaining the usage status of the load facility 5, a hotel (accommodation facility) is exemplified as the customer facility 3. The usage status of the hotel can be grasped based on the number of reservations made by users who reserve the hotel, the number of users checking into the hotel, and the connection status of the free Wi-Fi provided by the hotel. When there are many users using the hotel, the usage of the load facility 5 also increases, and the power consumption becomes large. Also, when there are few users using the hotel, the usage of the load facility 5 decreases, and the power consumption becomes small. Therefore, the information on the usage status of the customer facility 3 can be used as related information. Note that the hotel exemplified as the customer facility 3 is just an example and is not limited to this. Also, the usage status of the customer facility 3 can be grasped by obtaining the congestion information of the customer facility 3. Regarding the method of obtaining the congestion information, for example, it can be obtained using an online map service. Also, the information on the usage status of the load facility 5 installed in the customer facility 3 can be individually obtained and used as related information.

[0031] Each of the above information is an example of the information acquired by the acquisition unit 22. Among the above information acquired by the acquisition unit 22, for example, information that can be acquired on its own server, such as season, date, day of the week, time, etc., may be acquired by itself. Also, information such as weather information, event information of the region, and usage status of the load facility 5 can be acquired from, for example, the websites that provide each information. Log information on the power usage of customers can be acquired from, for example, the control device 4 and the load facility 5 installed in the customer facility 3. Note that these information may be prepared in advance by the user.

[0032] As described above, multiple pieces of related information can be listed. In reality, these pieces of information are combined and have a complex impact on the power consumption of the load equipment 5. Therefore, in this embodiment, a machine learning model showing the relationship between the related information and the power consumption of the load equipment 5 is used to predict the power consumption of the load equipment 5. The machine learning model is obtained by performing machine learning using the related information and the log information of the power consumption of the load equipment 5 corresponding to this related information. The machine learning model used in this embodiment outputs the probability distribution of the power consumption of the load equipment 5 based on the related information.

[0033] <Function of the generation unit> The generation unit 23 performs machine learning using the learning data and generates a machine learning model. The learning data refers to the information used for the learning of the machine learning model. Supervised learning is used for the learning of the machine learning model. In this embodiment, as the explanatory variables of the machine learning model, information that affects the power consumption of the load equipment 5 during past power consumption, that is, the related information, is used. Also, in this embodiment, as the objective variable of the machine learning model, the power consumption of the load equipment 5 during past power consumption is used.

[0034] <Function of the storage unit> The storage unit 24 stores various types of information. For example, the storage unit 24 stores the machine learning model in association with the customer facility 3. In the above, it was assumed that the generation unit 23 generates the machine learning model, but it may also be generated by an external system. In this case, the machine learning model generated in the external system is associated with the customer facility 3 and stored in the storage unit 24.

[0035] In addition, the memory unit 24 stores the threshold values used for the determination by the determination unit 26. The threshold values used for the determination include a peak cut threshold value and a determination reference value. The peak cut threshold value is a threshold value serving as a reference for restricting the power consumption of the load facility 5. The peak cut threshold value is set based on, for example, the value of the contract power of the consumer. More specifically, in consideration of the deviation and error between the instruction and the operation in the control of the power consumption of the load facility 5, a value slightly smaller than the contract power by a certain amount may be set as the peak cut threshold value. The determination reference value is a threshold value serving as a reference for determining whether to restrict the power consumption of the load facility 5. The determination reference value is specified based on the probability distribution of the power consumption amount predicted by the prediction unit 25. Details of the determination reference value will be described later.

[0036] <Function of the prediction unit> The prediction unit 25 predicts the probability distribution of the power consumption amount of the load facility 5 at the time of prediction by inputting the related information at the time of prediction execution into the machine learning model. In this embodiment, the related information input into the machine learning model is acquired at the time of prediction execution. The prediction unit 25 is an example of prediction means. Here, in this embodiment, the probability distribution of the power consumption amount predicted by the prediction unit 25 will be described with reference to a figure.

[0037] FIG. 4 is a diagram showing an example of the probability distribution of the power consumption amount. As shown in FIG. 4, the probability distribution represents the probability of taking each value with respect to the power consumption amount of the load facility 5, which is a random variable. In the graph shown in FIG. 4, the vertical axis represents the random variable, and the horizontal axis represents the probability density. In this embodiment, the random variable is the power consumption amount of the load facility 5. Also, the probability density represents the relative ease of occurrence of each value of the power consumption amount of the load facility 5, which is a random variable. The curve (A) shown in FIG. 4 represents the probability distribution of the predicted value of the power consumption amount predicted by the prediction unit 25 at a certain point in time. The probability distribution of the power consumption amount generally follows a normal distribution as shown by the curve (A) shown in FIG. 4.

[0038] <Function of the determination unit> The determination unit 26 determines whether to limit the power consumption of the load facility 5 at the time to be predicted, using the predicted probability distribution of the power consumption and a predetermined peak cut threshold value. The determination by the determination unit 26 will be described later.

[0039] <Function of the control unit> The control unit 27 controls the operations of each unit. Further, the control unit 27 of the present embodiment manages the control devices 4 of each customer facility 3 to be managed. Further, in the present embodiment, the control unit 27 outputs an instruction to limit the power consumption of the load facility 5 based on the determination of the determination unit 26. The above instruction is transmitted to the control device 4 of the customer facility 3 via the communication unit 21. Based on the instruction, in the customer facility 3, the power consumption of the load facility 5 is restricted by the control of the control device 4. The control unit 27 is an example of a control means.

[0040] <Control device> As described above, the control device 4 is installed in the customer facility 3. The control device 4 communicates with the load facility 5 installed in the customer facility 3 and manages the power consumption of individual devices, facilities, etc. Hereinafter, the control device 4 will be described.

[0041] <Hardware configuration of the control device> FIG. 5 is a diagram showing an example of the hardware configuration of the control device 4. The control device 4 is provided with an arithmetic processing unit 61, a storage device 62 for storing various information, and a communication interface 63 for communicating with an external device. The arithmetic processing unit 61 is configured by a computer. The arithmetic processing unit 61 has a CPU 61a as an example of a processor that executes various processes to be described later. Further, the arithmetic processing unit 61 has a ROM 61b in which a program is stored and a RAM 61c used as a work area. The storage device 62 is realized by an existing device such as a hard disk drive or a semiconductor memory. The arithmetic processing unit 61 and the storage device 62 are connected through a bus 64 and signal lines (not shown). Further, the control device 4 in the present embodiment may have a display device 65 and an input device 66.

[0042] The program executed by the CPU 61a can be provided to the control device 4 while being stored in a computer-readable recording medium such as a magnetic recording medium, an optical recording medium, a magneto-optical recording medium, or a semiconductor memory. Also, the program executed by the CPU 61a may be provided to the control device 4 using a communication means. The program provided to the control device 4 is stored in the storage device 62.

[0043] In this specification, the processor refers to a processor in a broad sense and includes a general-purpose processor and a dedicated processor. Also, the operation of the processor may be achieved not only by one processor but also by a plurality of physically separated processors cooperating with each other. Also, the order of each operation of the processor is not limited to the order described in this embodiment and may be changed.

[0044] <Functional Configuration of Control Device> FIG. 6 is a diagram showing a functional configuration example of the control device 4. The control device 4 in this embodiment includes a customer facility communication unit 71 and a customer facility control unit 72. These functions are realized by the cooperation of hardware and software. For example, the function of the customer facility communication unit 71 is realized by the communication interface 53. Also, the operation of the customer facility control unit 72 is performed by the CPU 61a executing a program stored in the storage device 62. Each function will be described below.

[0045] The consumer facility communication unit 71 communicates with the information processing device 2, the load equipment 5, etc., and transmits and receives various information and instructions. The consumer facility control unit 72 outputs a control command to the load equipment 5 and manages each device, each facility, etc. Further, the consumer facility control unit 72 receives an instruction from the information processing device 2 and restricts the power consumption of the load equipment 5. As shown in FIG. 1, in the consumer facility 3, as the load equipment 5, a load device 6, a storage battery 7, a charging facility 8, etc. are installed. These devices, facilities, etc. have different power consumption amounts respectively. Also, depending on the situation of the consumer facility 3, it is necessary to make the load equipment 5 to be the target of power consumption restriction different. The consumer facility control unit 72 may appropriately restrict the power consumption of the load equipment 5 by individually restricting the power consumption of these devices, facilities, etc. Also, by discharging the power stored in the storage battery 7 and using it as an alternative power source to the power grid 9, the power consumption in the consumer facility 3 may be suppressed.

[0046] <Generation of Machine Learning Model> A case of generating a machine learning model to be applied to the load equipment 5 in a certain consumer facility 3 will be described as an example. In the present embodiment, the prediction target is the average power consumption in a 30 - minute time unit according to the demand time limit. In the prediction, the prediction unit 25 obtains the average power consumption of the time unit including the prediction target time based on the related information acquired at the time of prediction execution. Therefore, using the related information at a certain past time point as an explanatory variable, the machine learning model is learned with the historical information of the average power consumption of the time unit including the time point corresponding to the time interval between the time of prediction execution and the prediction target time from that time point as the teacher data of the objective variable.

[0047] The time interval between the time of prediction execution and the prediction target time is preferably short. For example, any time point in the time unit immediately before the time unit including the prediction target time may be set as the time of prediction execution. In this case, using the related information at a certain past time point as an explanatory variable, the machine learning model is learned with the average power consumption in the next time unit of the time unit including that time point as the teacher data of the objective variable.

[0048] After machine learning is performed, through accuracy evaluation of the machine learning model and the like, a machine learning model is generated. The machine learning model is associated with the customer facility 3 and stored in the storage unit 24. This machine learning model is used when the prediction unit 25 predicts the power consumption of the load equipment 5.

[0049] <Prediction of Probability Distribution of Power Consumption> Next, the prediction of the probability distribution of the power consumption will be described. In this embodiment, the acquisition unit 22 acquires related information, and the prediction unit 25 predicts the probability distribution of the power consumption using the machine learning model. More specifically, the acquisition unit 22 acquires related information at the time of prediction execution when the prediction unit 25 makes a prediction. Then, the prediction unit 25 predicts the power consumption at the time of prediction target based on the acquired information, and outputs the prediction result as the probability distribution of the power consumption.

[0050] <Restriction on Power Consumption of Load Equipment> Next, the restriction on the power consumption of the load equipment 5 will be described. First, the determination by the determination unit 26 will be described. The determination unit 26 makes a determination using the predicted probability distribution of the power consumption, the peak cut threshold value, and the determination reference value. The determination reference value is a threshold value specified based on the predicted probability distribution of the power consumption as described above. More specifically, the determination reference value is a threshold value set for the probability that the predicted power consumption becomes a value lower than the peak cut threshold value in the predicted probability distribution of the power consumption of the load equipment 5. Note that in the probability distribution, the range in which the probability of a value satisfying a certain condition can be taken is 0 to 1. Therefore, the determination reference value can be arbitrarily determined between 0 and 1. Hereinafter, the determination will be described.

[0051] When making a determination, the determination unit 26 focuses on whether the probability that the predicted power consumption is less than the peak cut threshold value is greater than the determination reference value in the probability distribution of the power consumption predicted by the prediction unit 25. When the probability that the predicted power consumption is less than the peak cut threshold value is less than the determination reference value, the determination unit 26 determines to limit the power consumption of the load equipment 5. On the other hand, when the probability that the predicted power consumption is less than the peak cut threshold value is greater than or equal to the determination reference value, the determination unit 26 determines not to limit the power consumption of the load equipment 5. Hereinafter, specific examples will be given for explanation.

[0052] FIG. 7 is a diagram showing a first aspect of the probability distribution of power consumption at a certain point in time. In FIG. 7, the vertical axis represents the power consumption of the load equipment 5, and the horizontal axis represents the probability density. Also, the dotted line (B) in FIG. 7 represents the peak cut threshold value. The curve (C) in the figure represents the first aspect of the probability distribution of the power consumption predicted by the prediction unit 25 at a certain point in time. Also, among this curve (C), the power consumption with the largest probability density is represented by P1. In the figure shown in FIG. 7, the power consumption P1 exceeds the peak cut threshold value. Also, among this curve (C), the hatched area (D) represents the probability distribution of the power consumption predicted to be less than the peak cut threshold value.

[0053] Here, the case where an arbitrarily determinable determination reference value is set to 0.3, for example, will be described as an example. In FIG. 7, in the probability distribution of the power consumption predicted by the prediction unit 25, assume that the probability that the predicted power consumption is less than the peak cut threshold value, that is, the value of the hatched area (D) is calculated to be 0.2. In this case, the probability (0.2) of the hatched area (D) is less than the determination reference value (0.3). Therefore, the determination unit 26 determines to limit the power consumption of the load equipment 5.

[0054] However, the curve of the probability distribution does not necessarily look like that shown in the figure. When the predicted probability distribution is wide, it is also conceivable to obtain a determination result different from the determination result based on the probability distribution shown in FIG. 7.

[0055] FIG. 8 is a diagram showing a second mode of the probability distribution of power consumption at a certain point in time. In FIG. 8, a second mode of the probability distribution of the power consumption predicted by the prediction unit 25 at a certain point in time is shown by a curve (E). Comparing with FIG. 7, in FIG. 8, among the probability distributions indicated by the curve (E) in the figure, the power consumption P1 with the largest probability density is the same, but the distributions are different. As the probability distribution predicted by the prediction unit 25, a case where the distribution becomes wider as shown in FIG. 8 is also conceivable. And, among this curve (E), the hatched area (F) shows the probability distribution of the power consumption predicted to be below the peak cut threshold. The peak cut threshold and the determination reference value are set under the same conditions as those shown in FIG. 7.

[0056] In FIG. 8, in the probability distribution of the power consumption predicted by the prediction unit 25, assume that the probability that the predicted value of the power consumption becomes a value below the peak cut threshold, that is, the value of the hatched area (F) is calculated to be 0.4. In this case, the probability (0.4) of the hatched area (F) is equal to or greater than the determination reference value (0.3). Therefore, the determination unit 26 determines not to limit the power consumption of the load facility 5.

[0057] In the present embodiment, even when the power consumption P1 (see FIG. 8) is assumed to exceed the peak cut threshold, there may be a case where a determination result of not limiting the power consumption of the load facility 5 is obtained based on the predicted probability distribution.

[0058] In the above, when making a determination, the determination reference value is set to 0.3, but the determination reference value is a value that can be arbitrarily determined. Therefore, for example, taking the case where the determination reference value is set to 0.7 as an example, the following description will be given.

[0059] FIG. 9 is a diagram showing a third mode of the probability distribution of power consumption at a certain point in time. In FIG. 9, similar to FIG. 7, the vertical axis represents the power consumption of the load facility 5, and the horizontal axis represents the probability density. Also, the dotted line (B) in FIG. 9 represents the peak cut threshold. The curve (G) in the figure represents the probability distribution of the power consumption predicted by the prediction unit 25 at a certain point in time. Also, among this curve (G), the power consumption with the highest probability density is represented by P2. In the figure shown in FIG. 9, the power consumption P2 does not exceed the peak cut threshold. Also, among this curve (G), the hatched area (H) represents the probability distribution of the power consumption predicted to be below the peak cut threshold.

[0060] In FIG. 9, in the probability distribution of the power consumption predicted by the prediction unit 25, assume that the probability that the predicted power consumption becomes a value below the peak cut threshold, that is, the value of the hatched area (H) is calculated to be 0.8. In this case, the probability (0.8) of the hatched area (H) is equal to or greater than the determination reference value (0.7). Therefore, the determination unit 26 determines not to limit the power consumption of the load facility 5.

[0061] However, as described above, the curve of the probability distribution does not necessarily look like that shown in the figure. When the predicted probability distribution is wide, it is also conceivable to obtain a determination result different from the determination result based on the probability distribution shown in FIG. 9.

[0062] FIG. 10 is a diagram showing a fourth aspect of the probability distribution of the power consumption at a certain point in time. In FIG. 10, the fourth aspect of the probability distribution of the power consumption predicted by the prediction unit 25 at a certain point in time is shown by the curve (I). FIG. 10 shows that, compared with FIG. 9, among the probability distributions of the power consumption shown by the curve (I) in the figure, the power consumption P2 with the highest probability density is the same, but the distributions are different. It is also conceivable that the distribution becomes wider as shown in FIG. 10 as the probability distribution predicted by the prediction unit 25. And among this curve (I), the hatched area (J) represents the probability distribution of the power consumption predicted to be below the peak cut threshold. Note that the peak cut threshold and the determination reference value are set under the same conditions as those shown in FIG. 9.

[0063] In FIG. 10, in the probability distribution of the power consumption predicted by the prediction unit 25, assume that the probability that the predicted power consumption value is less than the peak cut threshold value, that is, the value of the shaded area (J) is calculated to be 0.6. In this case, the probability (0.6) of the shaded area (J) is less than the determination reference value (0.7). Therefore, the determination unit 26 determines that the power consumption of the load facility 5 is to be restricted.

[0064] As described above, by arbitrarily setting the determination reference value, even when the power consumption P2 (see FIG. 10) is not supposed to exceed the peak cut threshold value, there may be a case where a determination result of restricting the power consumption of the load facility 5 is obtained based on the predicted probability distribution.

[0065] <Aspects in the restriction of power consumption> Next, the aspect of restricting the power consumption of the load facility 5 will be described. When the determination unit 26 determines to restrict the power consumption of the load facility 5 based on the probability distribution of the power consumption predicted by the prediction unit 25, the control unit 27 outputs an instruction to restrict the power consumption of the load facility 5. The instruction is input to the control device 4 of the consumer facility 3 via the communication interfaces of each of the information processing device 2 and the consumer facility 3. Then, the consumer facility control unit 72 of the control device 4 issues instructions to each device, and the power consumption of the load facility 5 is restricted. Note that the restriction of the power consumption of the load facility 5 is performed by the CPU 61a of the control device 4 executing a program stored in the storage device 62.

[0066] As an example of the consumer facility 3, as described above, for example, houses, offices, factories, etc. can be mentioned. As a mode of restricting the power consumption of the load equipment 5 in these consumer facilities 3, for example, the mode of restricting the power consumption of the load device 6 shown in FIG. 1 can be mentioned. As the load device 6, for example, as described above, air conditioners, lighting, personal computers, etc. can be mentioned. Further, when the consumer facility 3 is a factory, in addition to the above loads, production equipment in a manufacturing factory, etc. can also be considered as the load device 6. By sending instructions to these devices and restricting their operation, it becomes possible to suppress power consumption. For example, in the case of an air conditioner, a mode of changing the set temperature can be considered. Also, in the case of lighting, the number of operating lights may be decreased. As another mode, as shown in FIG. 1, by supplementing the power used by the load equipment 5 with the power stored in the storage battery 7, the power supplied from the power grid 9 may be reduced. Furthermore, as another mode, as shown in FIG. 1, the power consumption of the charging equipment 8 may be restricted. Hereinafter, the mode of restricting the power consumption of the charging equipment 8 will be described.

[0067] As the charging equipment 8, as described above, for example, charging equipment for charging electric vehicles can be considered. When there are a plurality of charging equipment 8, among the plurality of charging equipment 8, by operating some and stopping some, power consumption may be suppressed. Note that the charging equipment 8 to be operated may be determined in advance.

[0068] Also, as an example of the charging equipment 8 for charging electric vehicles, there are normal chargers and rapid chargers. Rapid charging consumes more power than normal charging. Therefore, for example, by turning off the rapid charger and operating a specific normal charger, it is also possible to suppress power consumption compared to the case where the charging equipment 8 is fully operated.

[0069] On the other hand, in workplaces that frequently use electric vehicles, etc., if the charging by the charging facility 8 is restricted, it may cause problems in business operations. In such cases, charging by the charging facility 8 may be prioritized. In this case, it is conceivable to limit the power consumption of the load facility 5 different from the charging facility 8. For example, instead of the charging facility 8, the power consumption of the load device 6 is restricted. Also, the storage battery 7 may be used as a power source to replace the power grid 9, and the power received from the power grid 9 may be reduced.

[0070] Also, a charging amount for charging the electric vehicle may be determined in advance, and the target of restricting power consumption may be switched using this charging amount as a threshold value. For example, when the remaining battery level of the electric vehicle is less than the predetermined charging amount, the charging facility 8 is not restricted, and other load facilities 5 are restricted. Then, when the remaining battery level of the electric vehicle reaches the predetermined charging amount, the operation of the charging facility 8 is stopped. Thereby, while maintaining the charging amount required for the electric vehicle, the operation of the charging facility 8 can be restricted. Note that the charging amount for charging the electric vehicle can be arbitrarily determined. Settings including the charging amount, etc. may be stored in advance in the storage device 62 constituting the control device 4, for example.

[0071] Also, in the above, the modes such as the stop and switching of the operation of the charging facility 8 to be restricted have been described, but it is not limited to this. As a method of restricting power consumption, the charging power by the charging facility 8 may be determined. For example, when it is necessary to restrict power consumption, the power supply during charging by the charging facility 8 is decreased. Also, power consumption may be restricted by restricting the charging speed.

[0072] In the above description, the control device 4 manages the load facilities 5 individually. On the other hand, each device, facility, etc. of the load facilities 5 may have the functions of the control device 4. In this case, each device, facility, etc. communicates directly with the information processing device 2. As a result, compared with the case of managing the operation of each load facility 5 using the control device 4, even when the control device 4 fails, it is possible to prevent a situation where the control process stops functioning all at once. Also, when each device, facility, etc. has the functions of the control device 4, since the control process is performed directly between the information processing device 2 and each device, facility, etc., this embodiment is also applicable when using each device, facility, etc. outside the customer facility 3.

[0073] As described above, embodiments of the present invention have been explained, but the technical scope of the present invention is not limited to the above embodiments. For example, in this embodiment, as information that affects the power consumption of the load facilities 5, for example, season, date, day of the week, time, weather information, event information of the area to which the load facilities 5 belong, and the usage status of the load facilities 5 are listed, but it is not limited to the above. Any information that affects power consumption may be used other than the above.

[0074] Also, in this embodiment, the prediction unit 25 predicted the average power consumption per unit of time (for example, 30 minutes). On the other hand, instead of the average power consumption per unit of time, the power consumption at a certain point in time may be predicted. In this case, the prediction execution time may be set immediately before the time to be predicted. That is, relevant information is acquired and the power consumption is predicted immediately. When generating a machine learning model used for such prediction, for example, the relevant information at a certain point in the past may be used as an explanatory variable, and the power consumption at that time may be used as teacher data for the target variable for learning.

[0075] Also, in this embodiment, relevant information is collected at the time of prediction execution. Therefore, even after generating the machine learning model, the machine learning model may be further learned using the relevant information collected at the time of prediction execution and the actual power consumption identified at the time to be predicted. This can contribute to improving the accuracy of the machine learning model.

[0076] In addition, in the present embodiment, it is assumed that information affecting the power consumption of the load equipment 5 differs for each of the consumer facilities 3, and a machine learning model is generated for each consumer facility 3. On the other hand, considering that factors affecting power consumption differ depending on the use of the device, etc., a machine learning model may be generated for each device and equipment installed in the consumer facility 3. Also, considering that factors affecting power consumption are approximated in neighboring facilities, etc., one machine learning model may be shared among a plurality of consumer facilities 3. In addition, various changes and alternative configurations that do not depart from the scope of the technical idea of the present invention are included in the present invention.

Explanation of Signs

[0077] 1…System, 2…Information processing device, 3…Consumer facility, 4…Control device, 5…Load equipment, 6…Load device, 7…Storage battery, 8…Charging equipment, 9…Power grid, 21…Communication unit, 22…Acquisition unit, 23…Generation unit, 24…Storage unit, 25…Prediction unit, 26…Judgment unit, 27…Control unit

Claims

1. An acquisition unit that acquires related information, which is information defined in advance as information that affects the power consumption of a load device; A prediction unit that inputs the related information at the time of prediction execution into a machine learning model generated by performing machine learning using the related information at the time of past power consumption as an explanatory variable and the power consumption of the load device at the time of past power consumption as an objective variable, and predicts the probability distribution of the power consumption of the load device at the time of prediction target to which the result of the prediction is applied; A control unit that restricts the power consumption of the load device at the time of the prediction target based on the probability distribution of the power consumption of the load device at the time of the prediction target An information processing apparatus comprising the above.

2. The related information includes at least any one of season, date, day of the week, time, weather information, event information of the region to which the load device belongs, usage status of the facility to which the load device belongs, and usage status of the load device. The information processing apparatus according to claim 1, characterized by the above.

3. The acquisition unit acquires the related information at the time of prediction execution. The information processing apparatus according to claim 2, characterized by the above.

4. An acquisition means for acquiring related information, which is information defined in advance as information that affects the power consumption of a load device including a charging device; A prediction means for inputting the related information at the time of prediction execution into a machine learning model generated by performing machine learning using the related information at the time of past power consumption as an explanatory variable and the power consumption of the load device at the time of past power consumption as an objective variable, and predicting the probability distribution of the power consumption of the load device at the time of prediction target to which the result of the prediction is applied; A control means for determining the charging power of the charging device based on the probability distribution of the power consumption of the load device at the time of the prediction target A charging system comprising the above.

5. The control means operates the charging device defined in advance based on the probability distribution of the power consumption of the load device at the time of the prediction target. The charging system according to claim 4, characterized by the above.

6. When the remaining battery level of the electric vehicle charged by the charging device is less than a predetermined charging amount, the control means does not restrict the charging device. The charging system according to claim 4, characterized by the above.

7. A step of acquiring related information, which is information defined in advance as information that affects the power consumption of a load device; A step of predicting a probability distribution of the power consumption of the load equipment at the time of prediction target to which the result of the prediction is applied, by inputting the related information at the time of prediction execution for prediction into a machine learning model generated by performing machine learning with the related information at the time of past power consumption as an explanatory variable and the power consumption of the load equipment at the time of past power consumption as an objective variable; A step of restricting the power consumption of the load equipment at the time of prediction target based on the probability distribution of the power consumption of the load equipment at the time of prediction target; A power control method comprising:

8. A program for causing a computer to acquire related information which is information predetermined as information affecting the power consumption of the load equipment; predict a probability distribution of the power consumption of the load equipment at the time of prediction target to which the result of the prediction is applied, by inputting the related information at the time of prediction execution for prediction into a machine learning model generated by performing machine learning with the related information at the time of past power consumption as an explanatory variable and the power consumption of the load equipment at the time of past power consumption as an objective variable; restrict the power consumption of the load equipment at the time of prediction target based on the probability distribution of the power consumption of the load equipment at the time of prediction target. A program for realizing the above.

Images