Program, power demand forecasting method, and information processing device

Abstract

To improve the accuracy of predicting the amount of electricity demanded by a target consumer.SOLUTION: A machine learning model is generated by machine learning that uses calendar information indicating time and date at which electricity was consumed in at least one consumer and production result information of a product produced by the at least one consumer at the time and date, as an explanatory variable, and the amount of power consumed by the at least one consumer at the time and date, as an objective variable. A server 60 acquires calendar information indicating a prediction target period, acquires production schedule information of a product to be produced in the prediction target period by a target consumer that produces the product, and inputs the acquired calendar information and the production schedule information to the machine learning model, to predict the amount of electricity to be demanded by the target consumer in the prediction target period.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a technique for predicting power demand in a target consumer. [Background technology]

[0002] There are known techniques for predicting power demand. For example, Patent Document 1 describes a technique for generating a demand fluctuation pattern from actual demand data for past days that correspond to the same day of the week as the prediction day, and then linking the generated demand fluctuation pattern to actual values ​​at the current time to calculate a basic fluctuation pattern, which is a predicted pattern for power demand from the current time onwards. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2016-19358 Summary of the Invention [Problem to be solved by the invention]

[0004] The technology described in Patent Document 1 predicts power demand based on information such as the day of the week on which power is consumed. However, for consumers who produce products, there are other factors that affect the amount of power consumed in addition to information such as the day of the week. Therefore, when the technology described in Patent Document 1 is used to predict power demand for such consumers, the accuracy of the power demand prediction may be low.

[0005] One of the objects of the present invention is to improve the accuracy of predicting the amount of power demand in a target consumer. [Means for solving the problem]

[0006] One aspect of the present disclosure provides a program for causing a computer to execute the following steps: acquiring first calendar information indicating a prediction period; acquiring production schedule information for a target consumer that produces a product during the prediction period; and predicting the electricity demand of the target consumer during the prediction period by inputting the first calendar information and the production schedule information into a machine learning model generated by performing machine learning using second calendar information indicating the date and time at which electricity was consumed by at least one consumer and actual production information for the product at the at least one consumer at the date and time as explanatory variables, and the amount of electricity consumed by the at least one consumer at the date and time as a target variable.

[0007] Another aspect of the present disclosure provides an electricity demand forecasting method including the steps of acquiring first calendar information indicating a forecast period, acquiring production schedule information for a target consumer that produces a product during the forecast period, and predicting the electricity demand of the target consumer during the forecast period by inputting the first calendar information and the production schedule information into a machine learning model generated by performing machine learning using second calendar information indicating the date and time at which electricity was consumed by at least one consumer and actual production information for the product of the at least one consumer at the date and time as explanatory variables and the amount of electricity consumed by the at least one consumer at the date and time as a target variable.

[0008] Yet another aspect of the present disclosure provides an information processing device comprising: a first acquisition means for acquiring first calendar information indicating a prediction period; a second acquisition means for acquiring production schedule information for a target consumer that produces a product during the prediction period; and a prediction means for predicting the electricity demand of the target consumer during the prediction period by inputting the first calendar information and the production schedule information into a machine learning model generated by performing machine learning using second calendar information indicating the date and time at which electricity was consumed by at least one consumer and actual production information of the product for the at least one consumer at the date and time as explanatory variables, and the amount of electricity consumed by the at least one consumer at the date and time as a target variable. [Effects of the Invention]

[0009] According to the present invention, it is possible to improve the accuracy of predicting the amount of power demand in a target consumer. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a diagram showing an overview of a power control system according to an embodiment; [Figure 2] FIG. 2 is a diagram illustrating an example of the functional configuration of the power control system. [Figure 3] FIG. 2 is a diagram illustrating an example of a hardware configuration of a server. [Figure 4] 10 is a sequence chart illustrating an example of the operation of a power control system that generates a machine learning model. [Figure 5] 10 is a flowchart illustrating an example of the operation of a server that predicts the amount of power demand. DETAILED DESCRIPTION OF THE INVENTION

[0011] 1. Configuration FIG. 1 is a diagram illustrating an overview of a power control system 1 according to one embodiment. The power control system 1 is a so-called FEMS (Factory Energy Management System). The power control system 1 predicts the amount of power demand for each consumer H in order to optimize energy management at each consumer H. A consumer H refers to a person who receives and uses electricity. In one example, the consumer H is a specific consumer that receives high voltage or extra-high voltage electricity, specifically a facility such as a factory that produces products. The products may be chemicals, industrial products, or agricultural produce. The consumer H consumes electricity to produce the products. Therefore, the amount of power consumed by the consumer H varies depending on the production status of the products. The power control system 1 includes a power generation device 10, multiple power loads 20, a watt-hour meter 30, a measuring instrument 40, a control device 50, and a server 60 (an example of an information processing device).

[0012] Of these, the power generation plant 10, the multiple power loads 20, the watt-hour meter 30, the measuring instrument 40, and the control device 50 are installed indoors or outdoors at the consumer H. To simplify the drawing, only a single consumer H is shown in FIG. 1, but the power control system 1 may include multiple consumers H. Furthermore, each consumer H is also supplied with power from a commercial power source (omitted in FIG. 1). The power from the commercial power source is sold (or provided) by an electricity retailer. The server 60 is connected to the control device 50 of the consumer H via a network 90. ​​The control device 50 is connected to the power generation plant 10, the multiple power loads 20, the watt-hour meter 30, and the measuring instrument 40 via a network 95 such as a LAN (Local Area Network).

[0013] The power generation device 10 is a device that generates power. In one example, the power generation device 10 is a device that generates power using natural energy, specifically a solar power generation device or a wind power generation device. Alternatively, the power generation device 10 may be an emergency power generation device using a gas turbine or a diesel engine. The power load 20 is a device that consumes power at a consumer H. The power load 20 includes a storage battery 21 and a general power load 22.

[0014] The storage battery 21 (an example of a first power load) is a device that is charged using power supplied from the power generation device 10 or a commercial power source and discharges the stored power as needed. The operation of the storage battery 21 is controlled by the control device 50 based on an operation plan created by the server 60. The storage battery 21 has an upper limit on the amount of power consumption because charging operation stops when the storage battery 21 is fully charged. In one example, the storage battery 21 is an industrial storage battery, and specific examples include a NAS battery (registered trademark), a lithium-ion battery, a lead battery, and a nickel-metal hydride battery.

[0015] The general power load 22 (an example of a second power load) is a power load 20 other than the storage battery 21 installed in the consumer H, and includes machines and equipment used in the production of products, air conditioning and heating equipment, lighting equipment, etc. The operation of the general power load 22 is not controlled by the control device 50, and operates according to, for example, user operation or a PLC (Programmable Logic Controller).

[0016] The watt-hour meter 30 is a device that measures the total amount of power consumed by the consumer H. The total amount of power consumed by the consumer H is the sum of the amount of power consumed by the storage battery 21 and the amount of power consumed by the general power load 22. In one example, the watt-hour meter 30 is a so-called smart meter. The watt-hour meter 30 supplies the measured total amount of power consumed to the control device 50.

[0017] The measuring instrument 40 is a device that measures the amount of power consumed by the storage battery 21. The amount of power consumed by the storage battery 21 is, for example, the amount of power consumed by the storage battery 21. In one example, when the amount of power consumed per unit time of the storage battery 21 is approximately constant, the measuring instrument 40 measures the amount of power consumed by the storage battery 21 using the operating time of the storage battery 21 and the amount of power consumed per unit time. The measuring instrument 40 supplies the measured amount of power consumed by the storage battery 21 to the control device 50.

[0018] In this example, the measuring instrument 40 measures the amount of power consumption, but the measuring instrument 40 may measure the operating time, and the control device 50 may calculate the amount of power consumption using the operating time measured by the measuring instrument 40 and the amount of power consumption per unit time.

[0019] The control device 50 controls the power generation device 10, the power load 20, the watt-hour meter 30, and the measuring instrument 40 at the consumer H. For example, the control device 50 controls the operation of the storage battery 21 in accordance with instructions from the server 60. The control device 50 also transmits a history of the amount of power consumed at the consumer H to the server 60. The server 60 controls the power control system 1. For example, the server 60 creates an operation plan for the storage battery 21 and instructs the control device 50 to operate the storage battery 21 in accordance with this operation plan. The server 60 also predicts the amount of power demand at the consumer H based on the history of the amount of power consumed at the consumer H received from the control device 50.

[0020] 2 is a diagram illustrating an example of the functional configuration of the power control system 1. The power control system 1 has a storage unit 51, a transmission unit 52, a storage unit 61, an acquisition unit 62, an acquisition unit 63, an acquisition unit 64, an acquisition unit 65, a generation unit 66, and a prediction unit 67. Of these functional elements, the storage unit 51 and the transmission unit 52 are implemented in the control device 50. The storage unit 61, the acquisition unit 62, the acquisition unit 63, the acquisition unit 64, the acquisition unit 65, the generation unit 66, and the prediction unit 67 are implemented in the server 60.

[0021] In the control device 50, the storage means 51 stores a history of the total amount of power consumption at the consumer H measured by the watt-hour meter 30, and a history of the amount of power consumption at the storage battery 21 measured by the measuring instrument 40. Each history includes the date and time when power was consumed (hereinafter also referred to as the "power consumption date and time") and the amount of power consumption at the power consumption date and time. The transmission means 52 transmits the history of the total amount of power consumption at the consumer H and the history of the amount of power consumption at the storage battery 21 stored in the storage means 51 to the server 60.

[0022] In the server 60, the storage means 61 stores various data including a consumer database and calendar information. The consumer database stores various information related to each consumer H. The consumer database is used, for example, to identify the region to which the consumer H belongs and the input terminal (not shown) installed at the consumer H. In one example, the consumer database stores identification information for each consumer H, the address of the consumer H, and the address of the input terminal installed at the consumer H, in association with each other. The calendar information is, for example, information about this year's calendar. The calendar information includes, for example, month, day, day of the week, weekday, holiday, public holiday, beginning of the month, middle of the month, end of the month, number of weeks, season, sunrise, sunset, and major events. Weekday, holiday, and public holiday are examples of types of whether or not a day is a public holiday. Public holiday indicates that a day is a public holiday, and weekdays and holidays indicate that they are not public holidays. The beginning, middle, and end of the month are examples of types of whether or not a day is the end of the month. The end of the month indicates that it is the end of the month, and the beginning and middle of the month indicate that it is not the end of the month.

[0023] The acquisition means 62 acquires a history of total power consumption in each consumer H and a history of power consumption in the storage battery 21. The acquisition means 63 (an example of a first acquisition means) acquires calendar information indicating target dates and times. This calendar information includes calendar information indicating the dates and times of power consumption in the consumer H (an example of second calendar information) and calendar information indicating a prediction target period (an example of first calendar information). The prediction target period is a period that serves as a unit of prediction. Each piece of calendar information includes, for example, at least one of month, day, time, day of the week, weekday, holiday, public holiday, beginning of the month, middle of the month, end of the month, number of weeks, season, sunrise, sunset, and major event. In one example, the acquisition means 63 acquires the dates and times of power consumption included in each history acquired by the acquisition means 62. The acquisition means 63 also acquires additional calendar information indicating the dates and times of power consumption from the storage means 61. The additional calendar information includes calendar information other than date and time, such as at least one of day of the week, weekday, holiday, public holiday, beginning of the month, middle of the month, end of the month, number of weeks, season, sunrise, sunset, and major events. Furthermore, the acquisition means 63 acquires a prediction target period. Furthermore, the acquisition means 63 acquires additional calendar information indicating the prediction target period from the storage means 61.

[0024] The acquisition means 64 acquires weather information for the area to which each consumer H belongs at a target date and time. The weather information includes weather information for the area to which the consumer H belongs at the date and time of power consumption (an example of second weather information) and weather information for the area to which the consumer H belongs during the prediction target period (an example of first weather information). Note that since the prediction target period is a future period, the weather information for the prediction target period does not indicate actual weather, but indicates a weather forecast. Each piece of weather information includes, for example, at least one of weather, temperature, humidity, precipitation, solar radiation, wind direction, and wind speed. In one example, the acquisition means 64 acquires weather information for the area to which the consumer H belongs at the date and time of power consumption included in the history acquired by the acquisition means 62 from an external server operated by a business or organization that provides weather information. The acquisition means 64 also acquires weather information for the area to which the consumer H belongs during the prediction target period from an external server operated by a business or organization that provides weather information.

[0025] The acquisition means 65 (an example of a second acquisition means) acquires production information of each consumer H's product at a target date and time. The production information is information indicating the production status of the product. The production information includes actual production information of the product at the date and time of the consumer H's power consumption and production schedule information of the consumer H's product for the prediction target period. Each piece of production information includes at least one of the product type, production volume of the product, and production process of the product. If the product is produced through multiple production processes, this production process is the production process among these production processes that is actually performed at the target date and time. In one example, the consumer H creates actual production information and production schedule information and inputs them into an input terminal installed at the consumer H. The acquisition means 65 acquires the actual production information and production schedule information from this input terminal. Alternatively, the consumer H creates actual production information and production schedule information and provides them to the administrator of the server 60. The acquisition means 65 acquires actual production information and production schedule information input into the server 60 by the administrator's operation.

[0026] The generation means 66 generates a machine learning model for each consumer H by machine learning training data, where the objective variable is the amount of power consumed by the general power load 22 in each consumer H, and the explanatory variables are calendar information indicating the date and time of power consumption for that consumer H, weather information for the date and time of power consumption, and production record information acquired by the acquisition means 63 to 65. The generation means 66 uses the amount of power consumed by the general power load 22, obtained by subtracting the amount of power consumed by the storage battery 21 from the total amount of power consumed by the consumer H acquired by the acquisition means 62, as the objective variable. The generation means 66 then learns the relationship between the amount of power consumed by the general power load 22 in each consumer H and the calendar information indicating the date and time of power consumption for that consumer H, weather information for the date and time of power consumption, and production record information. Through this machine learning, a machine learning model is generated that outputs a prediction of the amount of power demand for the general power load 22 in the consumer H, using the calendar information, weather information, and production information as inputs. The power demand of the general power loads 22 is the amount of power predicted to be consumed by the general power loads 22 excluding the storage battery 21 from among the power loads 20 installed in the consumer H.

[0027] The reason for using the power consumption of the general power load 22 as the objective variable is to improve the accuracy of the power demand prediction. As described above, the operation of the storage battery 21 is controlled based on the operation plan created by the server 60. Because the power demand of the storage battery 21 is the amount of power consumed by the operation based on this operation plan, prediction based on the actual operation plan provides higher accuracy than prediction based on machine learning. Therefore, the generation means 66 performs machine learning using the power consumption of the general power load 22 as the objective variable to generate a machine learning model that predicts the power demand of the general power load 22 excluding the power demand of the storage battery 21. This makes it possible, for example, to predict the power demand of the general power load 22 using machine learning, and to separately predict the power demand of the storage battery 21 based on the operation plan. As a result, the accuracy of the power demand prediction for the consumer H is improved.

[0028] The reason why calendar information indicating the date and time of power consumption, weather information for the date and time of power consumption, and production performance information are used as explanatory variables is because these pieces of information affect the amount of power consumed by the general power load 22. For example, the amount of power consumed by the general power load 22 differs depending on the month, day, time, day of the week, whether it is a public holiday, whether it is the end of the month, etc. The amount of power consumed by the general power load 22 also differs depending on the weather, temperature, etc. Furthermore, the amount of power consumed by the general power load 22 differs depending on the production volume of products produced at consumer H, the type of product, and the production process for producing the product. Therefore, by performing machine learning using calendar information indicating the date and time of power consumption, weather information for the date and time of power consumption, and production performance information as explanatory variables, the accuracy of prediction of the power demand of the general power load 22 is improved.

[0029] The prediction means 67 inputs the calendar information indicating the prediction period, the second weather information, and the production schedule information acquired by the acquisition means 63 to 65 into the machine learning model for each consumer H generated by the generation means 66, and predicts the power demand for the prediction period for each consumer H. However, the power demand of the consumer H predicted by the prediction means 67 does not include the amount of power consumed by the storage battery 21.

[0030] FIG. 3 is a diagram illustrating an example of the hardware configuration of the server 60. The server 60 is a computer device having a CPU (Central Processing Unit) 601, memory 602, storage 603, and a communication IF (Interface) 604. The CPU 601 is a control device that executes programs to perform various calculations and controls other hardware elements of the server 60. The memory 602 is a main storage device that functions as a work area when the CPU 601 executes programs. The storage 603 is a non-volatile auxiliary storage device that stores various programs and data. The communication IF 604 is a communication device that communicates with other devices in accordance with a predetermined communication standard (e.g., Ethernet (registered trademark)). Note that when production performance information and production schedule information are input into the server 60 by an administrator, the server 60 further has input devices such as a mouse and a keyboard.

[0031] In this example, the storage 603 stores a program (hereinafter referred to as the "server program") for causing the computer device to function as the server 60 in the power control system 1. The CPU 601 executes the server program, thereby implementing the functions of FIG. 2 in the computer device. When the CPU 601 is executing the server program, at least one of the memory 602 and the storage 603 is an example of a storage means 61. The CPU 601 is also an example of acquisition means 62 to 65, generation means 66, and prediction means 67.

[0032] Although detailed illustration is omitted, the control device 50 includes a microcomputer including a processor, memory, storage, and a communication IF. This storage stores a program (hereinafter referred to as a "control program") for causing the microcomputer to function as the control device 50 in the power control system 1. The processor executes the control program, thereby implementing the functions shown in FIG. 2 in the microcomputer. When the processor is executing the control program, at least one of the memory and the storage is an example of storage means 51. The communication IF is an example of transmission means 52.

[0033] 2.Operation 2.1 Generating a Machine Learning Model 4 is a sequence chart illustrating the operation of the power control system 1 that generates a machine learning model according to one embodiment. The sequence in FIG. 4 is started when a reference time such as midnight arrives. In the following, functional elements such as the acquisition means 62 are described as the subject of processing, which means that a hardware element such as the CPU 601 that is executing a program such as a server program executes processing in cooperation with other hardware elements.

[0034] Prior to this process, the server 60 determines a target consumer H (hereinafter referred to as a "target consumer") from among the multiple consumers H included in the power control system 1. The target consumers are determined one by one in order according to a predetermined rule. Here, it is assumed that consumer H with identification information "001" is determined as the target consumer.

[0035] In step S101, the transmission means 52 of the control device 50 of the target consumer transmits the history of total power consumption stored in the storage means 51 and the history of power consumption of the storage battery 21 to the server 60 together with identification information of the control device 50 (or the target consumer) that is the transmission source. Each history indicates the power consumption in 30-minute increments, for example. Each history includes the date and time when the power was consumed and the power consumption amount at that date and time. The acquisition means 62 of the server 60 receives these histories and identification information from the control device 50.

[0036] In step S102, the acquisition means 63 acquires calendar information indicating the date and time of power consumption included in the history acquired in step S101. Here, it is assumed that the date and time of power consumption included in the history of power consumption amount for the target consumer is from 0:00 to 24:00 on December 1, 20XX. For example, the acquisition means 63 acquires the date and time of power consumption from 0:00 to 24:00 on December 1, 20XX as calendar information from the history acquired in step S101. The acquisition means 63 also acquires additional calendar information from the storage means 61 indicating the day of the week on December 1, 20XX, the holiday if this day is a national holiday, the end of the month if this day is the last day of the month, etc.

[0037] In step S103, the acquisition means 64 acquires weather information for the region to which the target consumer belongs at the date and time of power consumption included in the history acquired in step S101. For example, the acquisition means 64 identifies an address associated with identification information "001" in the consumer database stored in the storage means 61. Here, it is assumed that this address includes X City. In this case, the acquisition means 64 acquires weather information indicating the weather, temperature, etc. in X City from 0:00 to 24:00 on December 1, 20XX, from an external server.

[0038] In step S104, the acquisition means 65 acquires production performance information of the target consumer's products at the date and time of power consumption included in the history acquired in step S101. For example, the acquisition means 65 identifies the address of the input terminal associated with the identification information "001" in the consumer database stored in the storage means 61. The acquisition means 65 uses the identified address to acquire production performance information of the target consumer's products from 0:00 to 24:00 on December 1, 20XX from the input terminal installed in the target consumer. The production performance information includes the type of product produced at consumer H at the date and time of power consumption, the production volume, and the production process of the product performed at that date and time.

[0039] In step S105, the generation means 66 calculates the amount of power consumption of the general power load 22 in each time period by subtracting the amount of power consumption of the storage battery 21 from the total amount of power consumption in each time period included in the history acquired in step S101. For example, assume that the total amount of power consumption in the time period from 0:00 to 0:30 on December 1, 20XX is 10 kW and the amount of power consumption of the storage battery 21 is 5 kW. In this case, the amount of power consumption of the general power load 22 in this time period is 10 kW - 5 kW = 5 kW. For other time periods, the amount of power consumption of the general power load 22 is calculated in a similar manner.

[0040] In step S106, the generating means 66 generates a machine learning model by machine learning training data using the power consumption calculated in step S105 as the objective variable and the calendar information, weather information, and production performance information acquired in steps S102 to S104 as explanatory variables. The generating means 66 performs machine learning on the relationship between the power consumption and calendar information of the target consumer. Therefore, for example, if power consumption tends to be high in certain months and low in other months, the generating means 66 learns this trend. For example, if power consumption tends to be high at the end of the month and low in other periods, the generating means 66 learns this trend. For example, if power consumption tends to be high on days of the week when products are produced at the target consumer and low on other days of the week, the generating means 66 learns this trend. For example, if power consumption tends to be high on weekdays and low on weekends and holidays, the generating means 66 learns this trend. Whether or not it is the end of the month is used as an explanatory variable because the power consumption trend may differ between the end of the month and other days, even for the same day of the week. Whether or not a day is a public holiday is used as an explanatory variable because there is a possibility that the tendency of power consumption will differ between public holidays and weekdays even if the day is the same day of the week.

[0041] Furthermore, the generating means 66 performs machine learning on the relationship between the amount of power consumption and weather information for the target consumer. Therefore, for example, if there is a tendency for power consumption to be low on sunny days and high on cloudy and rainy days, the generating means 66 learns this tendency. For example, if there is a tendency for power consumption to be high on days with high and low temperatures and low on days with medium temperatures, the generating means 66 learns this tendency.

[0042] Furthermore, the generation means 66 performs machine learning on the relationship between the amount of power consumption and production performance information for the target consumer. Therefore, for example, if there is a tendency for power consumption to be high when production volume is high and low when production volume is low, the generation means 66 learns this tendency. For example, if there is a tendency for power consumption to be high when a specific type of product is produced and low when other types of product are produced, the generation means 66 learns this tendency. For example, if there is a tendency for power consumption to be high when a specific production process for producing a product is performed and low when other production processes are performed, the generation means 66 learns this tendency. Furthermore, if there is a tendency for power consumption to fluctuate depending on a combination of at least two of calendar information, weather information, and production performance information, the generation means 66 may learn this tendency. The generation means 66 stores the machine learning model thus generated in the storage means 61 in association with the identification information of the target consumer.

[0043] The server 60 repeatedly executes the processes of steps S101 to S106 while sequentially changing the target consumer. In this way, machine learning models are generated for all consumers H belonging to the power control system 1. Furthermore, even after machine learning models have been generated for all consumers H, the processes of steps S101 to S106 may be repeatedly executed. In this case, in step S106, further machine learning is performed using newly acquired data.

[0044] 2.2 Electricity demand forecast Fig. 5 is a flowchart illustrating the operation of the server 60 that predicts the amount of power demand according to one embodiment. The flow in Fig. 5 is started when a reference time such as midnight is reached. In the following, functional elements such as the acquisition means 62 are described as the subject of processing, which means that a hardware element such as the CPU 601 that is executing a program such as a server program executes processing in cooperation with other hardware elements.

[0045] Prior to this process, the server 60 determines a target consumer H (hereinafter referred to as a "target consumer") from among the multiple consumers H included in the power control system 1. The target consumers are determined one by one in order according to a predetermined rule. Here, it is assumed that consumer H with identification information "001" is determined as the target consumer.

[0046] In step S111, the prediction means 67 sets a prediction period. For example, when predicting the power demand for 24 hours from midnight on the next day at midnight every day, the prediction means 67 sets the prediction period from midnight to midnight on the next day. Here, the prediction period is assumed to be from midnight to midnight on December 14, 20XX.

[0047] In step S112, the acquisition means 63 acquires calendar information indicating the prediction target period set in step S111. For example, the acquisition means 63 acquires, as calendar information, the date and time indicating the prediction target period of December 14, 20XX, from 0:00 to 24:00, set in step S111. The acquisition means 63 also acquires, from the storage means 61, additional calendar information indicating the day of the week of December 14, 20XX, the holiday if this day is a national holiday, the end of the month if this day is the last day of the month, etc.

[0048] In step S113, the acquisition means 64 acquires weather information for the region to which the target consumer belongs for the prediction target period set in step S112. For example, the acquisition means 64 identifies an address associated with identification information "001" in the consumer database stored in the storage means 61. Here, it is assumed that this address includes X City. In this case, the acquisition means 64 acquires weather information indicating the weather, temperature, etc. in X City from 0:00 to 24:00 on December 14, 20XX, from an external server.

[0049] In step S114, the acquisition means 65 acquires production schedule information for the target consumer's product during the forecast period. For example, the acquisition means 65 identifies the address of the input terminal associated with the identification information "001" in the consumer database stored in the storage means 61. The acquisition means 65 uses the identified address to acquire production schedule information for the target consumer's product from 0:00 to 24:00 on December 14, 20XX, from the input terminal installed at the target consumer. The production schedule information includes the type of product scheduled to be produced during the forecast period, its production volume, and the production process for the product scheduled to be performed during the forecast period.

[0050] In step S115 (an example of a prediction step), the prediction means 67 inputs the calendar information, weather information, and production schedule information acquired in steps S112 to S114 into a machine learning model of the target consumer stored in the storage means 61, and predicts the power demand of the target consumer for the prediction target period. For example, the prediction means 67 uses a machine learning model stored in the storage means 61 in association with identification information "001." The prediction means 67 inputs the calendar information, weather information, and production schedule information acquired in steps S112 to S114 into this machine learning model, and predicts the power demand of the target consumer from 0:00 to 24:00 on December 14, 20XX, in 30-minute increments. However, this power demand prediction is a power demand prediction of the general power load 22 excluding the storage battery 21. In other words, this power demand prediction does not include a power demand prediction of the storage battery 21.

[0051] The server 60 repeatedly executes the processes of steps S111 to S115 while sequentially changing the target consumer. In this way, the power demand amount for the prediction period is predicted for all consumers H belonging to the power control system 1.

[0052] The consumer H uses the predicted power demand for the consumer H obtained in this manner to prepare the power generation device 10 and the storage battery 21 according to this power demand. When the power demand in society is high or during a disaster, the consumer H can supply the power generated by the power generation device 10 or charged in the storage battery 21 to the surrounding area. By supplying power to society in this manner, the consumer H can obtain benefits such as money. However, in order to supply power to society, the consumer H needs to generate and store more power than the amount of power consumed by the consumer H. Therefore, the consumer H calculates the required amount of power generation and storage based on the predicted power demand for the consumer H, and installs the power generation device 10 with the required amount of power generation and the storage battery 21 with the required amount of power storage at the consumer H. For example, if the amount of power generation or storage is insufficient, the consumer H may add more power generation devices 10 or storage batteries 21 to cover the shortfall.

[0053] Furthermore, the server 60 creates an operation plan for the storage battery 21 so as to optimize energy management. This operation plan includes an operation start time and an operation end time for the storage battery 21. The server 60 transmits an operation command for the storage battery 21 to the control device 50 according to the created operation plan. The control device 50 controls the operation of the storage battery 21 according to the command received from the server 60. If the amount of power consumed per unit time of the storage battery 21 is approximately constant, the server 60 can accurately predict the power demand of the storage battery 21 using the amount of power consumed per unit time and the operation plan of the storage battery 21. If the power demands of all the power loads 20, including the storage battery 21, were predicted by machine learning, the prediction accuracy would be lower for at least the power demand of the storage battery 21 than if the power demand were predicted based on the operation plan. As described above, by predicting the power demand of the general power loads 22 excluding the storage battery 21 at the consumer H by machine learning, the accuracy of the power demand prediction is higher than if the power demand including the power demand of the storage battery 21 were predicted by machine learning.

[0054] According to the embodiment described above, because production information of products that affect the amount of power consumed at the consumer H is included in the explanatory variables for machine learning, the accuracy of the power demand prediction is higher than when the explanatory variables do not include production information. Furthermore, because calendar information and weather information that affect the amount of power consumed at the consumer H are included in the explanatory variables for machine learning, the accuracy of the power demand prediction is higher than when the explanatory variables do not include this information. Furthermore, because the power demand of the general power loads 22 excluding the storage battery 21, the operation of which is controlled by the control device 50, among the power loads 20 installed at the consumer H, is predicted by machine learning, the accuracy of the power demand prediction is higher than when the power demand including the power demand of the storage battery 21 is predicted by machine learning. Furthermore, because the power demand of the general power loads 22 at the consumer H is predicted collectively by machine learning, the amount of calculation required for the prediction is smaller than when the power demand is predicted separately for each general power load 22.

[0055] 3. Variations The present invention is not limited to the above-described embodiment, and various modifications are possible. Several modifications will be described below. Two or more of the following modifications may be used in combination.

[0056] Variation 1 The production information is not limited to the type of product, its production volume, and the production process of the product. The production information may be any information that indicates the production status of the product. Furthermore, the explanatory variables used in machine learning do not necessarily include all of the calendar information, weather information, and production information. For example, one or two of the calendar information, weather information, and production information may be used as explanatory variables.

[0057] Variation 2 The explanatory variables used in machine learning are not limited to those exemplified in the above-described embodiments. The explanatory variables may be any information that affects the amount of power consumption at the consumer H. For example, the explanatory variables may include situation information indicating the situation of the consumer H. The situation information may include at least one of the above-described production information, the number of people in the consumer H, economic indicators related to the consumer H, such as stock prices and exchange rates, raw material prices, and disaster information. For example, when employees working at the consumer H punch their work start and end times using time cards, the number of people in the consumer H is calculated for each predetermined time period using the work start and end times. The economic indicators are provided, for example, by a business or organization that provides economic indicators. The prices of raw materials are provided, for example, by a party or organization that provides raw material prices. The disaster information is provided, for example, by an external server operated by a business or organization that provides weather information.

[0058] The acquisition means 65 acquires status information of the consumer H at a target date and time. This status information includes status information at the date and time of power consumption at the consumer H and status information for the prediction target period. Since the prediction target period is a future period, the status information for the prediction target period does not indicate actual results but indicates plans or predictions. In one example, the consumer H inputs status information to an input terminal installed at the consumer H. The acquisition means 65 acquires status information at the date and time of power consumption and status information for the prediction target period from this input terminal or an external server. The generation means 66 further performs machine learning using the status information at the date and time of power consumption acquired by the acquisition means 65 as an explanatory variable.

[0059] For example, if the amount of power consumed tends to be high when the number of people in consumer H is large and low when the number of people is small, the generation means 66 learns this trend. For example, if the amount of power consumed tends to be high when economic indicators indicate a good economy and low when economic indicators indicate a bad economy, the generation means 66 learns this trend. For example, if the amount of power consumed tends to be high when the prices of raw materials are low and low when the prices of raw materials are high, the generation means 66 learns this trend. For example, if the amount of power consumed tends to be high when a disaster occurs and low when no disaster occurs, the generation means 66 learns this trend.

[0060] The prediction means 67 further predicts the power demand by inputting the situation information for the prediction period acquired by the acquisition means 65 into the machine learning model. According to this modification, the situation information of consumer H that affects the power consumption amount at consumer H is used as an explanatory variable, thereby improving the accuracy of the power demand prediction by machine learning.

[0061] Furthermore, in this modification, only the status information correlated with the power consumption of the consumer H may be used from the status information acquired by the acquisition means 65. Here, it is assumed that multiple status information are acquired from the consumer H. The generation means 66 extracts status information correlated with the power consumption of the consumer H from the multiple status information using a known method, and uses the extracted status information as an explanatory variable. In one example, the generation means 66 calculates a correlation coefficient between each status information acquired by the acquisition means 65 and the power consumption of the consumer H. The generation means 66 extracts status information whose correlation coefficient is equal to or greater than a threshold from the multiple status information, and uses the extracted status information as an explanatory variable. The generation means 66 does not use status information whose correlation coefficient is less than the threshold as an explanatory variable. This threshold is preferably a positive value. According to this modification, only the status information correlated with the power consumption of the consumer H is used as an explanatory variable, thereby improving the accuracy of the power demand prediction using machine learning.

[0062] Variation 3 Machine learning for predicting the power demand of a consumer H does not necessarily need to use information about the consumer H itself. For example, if the consumer H has a small power consumption history, the power consumption history of another consumer H similar to the consumer H may be used as the power consumption history of the consumer H. For example, if production information of a product for the consumer H cannot be obtained, the production information of a product for another consumer H similar to the consumer H may be used as the production information for the product for the consumer H. It is preferable that the other consumer H is a consumer H whose profile, such as the organization to which the consumer belongs, the product, production volume, size, number of employees, etc., is similar to that of the consumer H. In one example, if multiple consumers H belong to a certain organization, a machine learning model generated by performing machine learning using the power consumption history and product production information of one consumer H belonging to this organization may be used to predict the power demand of other consumers H belonging to this organization. In short, at least one consumer H providing data used for machine learning may be the target consumer itself, a set of the target consumer and other consumers H, or one or more other consumers H not including the target consumer. The configuration according to this modification also makes it possible to predict the amount of power demand in the customer H.

[0063] Variation 4 Any learning algorithm may be used for machine learning, including, for example, a decision tree, a support vector machine, a neural network, and deep learning.

[0064] Variation 5 The power loads 20 whose operation is controlled by the control device 50 are not limited to the storage battery 21. The power loads 20 whose operation is controlled by the control device 50 may be any power loads 20 as long as the control device 50 can control the operation of the power loads 20 based on an operation plan created by the server 60 and the amount of power consumption can be measured individually. However, it is preferable that the power loads 20 whose operation is controlled by the control device 50 have an upper limit on the amount of power consumption.

[0065] Variation 6 The correspondence between the functional elements and the hardware elements in the power control system 1 is not limited to that illustrated in the embodiment. For example, some of the functions described as the server 60 in the embodiment may be implemented in the control device 50. Alternatively, some of the functions described as the server 60 in the embodiment may be implemented in another device on the network. In one example, another device on the network different from the server 60 may have a storage means 61, acquisition means 62 to 65, and generation means 66, and generate a machine learning model. The server 60 may have a storage means 61, acquisition means 62 to 65, and prediction means 67, acquire a machine learning model from the other device, and predict power demand using the acquired machine learning model. In this case, the server 60 and the other device may be managed and operated by different operators.

[0066] Variation 7 The operation of the power control system 1 is not limited to the above example. The order of the processing steps of the power control system 1 may be changed as long as there is no contradiction. Furthermore, some of the processing steps of the power control system 1 may be omitted.

[0067] Variation 8 Another aspect of the present invention may provide a power demand forecasting method having processing steps performed in at least one of the power control system 1, the power generation device 10, the power load 20, the electricity meter 30, the measuring instrument 40, the control device 50, and the server 60.

[0068] Variation 9 The various programs exemplified in the embodiments may be provided by downloading via a network such as the Internet, or may be provided in a state recorded on a recording medium such as a CD-ROM (Compact Disc Read Only Memory). [Explanation of symbols]

[0069] 1: Power control system, 10: Power generation device, 20: Power load, 21: Storage battery, 22: General power load, 30: Watt-hour meter, 40: Measuring instrument, 50: Control device, 51: Storage means, 52: Transmission means, 60: Server, 61: Storage means, 62: Acquisition means, 63: Acquisition means, 64: Acquisition means, 65: Acquisition means, 66: Generation means, 67: Prediction means, 90: Network, 95: Network, 601: CPU, 602: Memory, 603: Storage, 604: Communication IF

Claims

1. On the computer, acquiring first calendar information indicating a prediction target period and including at least one of a day of the week, a type of whether it is a public holiday, and a type of whether it is the end of the month, in the prediction target period; acquiring production schedule information of a target consumer that produces a product for the product during the forecast target period; a prediction step of predicting the amount of electricity demand of the target consumer during the prediction period by inputting the first calendar information and the production schedule information into a machine learning model generated by performing machine learning using second calendar information indicating a date and time when electricity was consumed in at least one consumer, the second calendar information including at least one of the day of the week, whether it is a public holiday, and whether it is the end of the month at the date and time, and production performance information of the at least one consumer at the date and time as explanatory variables, and the amount of electricity consumed by the at least one consumer at the date and time as a target variable; A program for executing a first power load, a second power load, and a control device that controls the first power load are installed in the at least one consumer; The objective variable is the power consumption amount obtained by subtracting the power consumption amount of the first power load from the sum of the power consumption amount of the first power load and the power consumption amount of the second power load at the date and time. program.

2. Each of the production performance information and the production schedule information includes at least one of the production amount of the product, the type of the product, and the production process of the product. The program according to claim 1.

3. a program for further executing a step of acquiring first meteorological information indicating weather in a region to which the target consumer belongs during the prediction target period, the explanatory variables include second weather information indicating weather at the date and time in a region to which the at least one consumer belongs, In the prediction step, the first weather information is further input to the machine learning model to predict the amount of electricity demand. The program according to claim 1 or 2.

4. The first power load includes a storage battery. The program according to claim 1.

5. A computer comprising: acquiring first calendar information indicating a prediction target period and including at least one of a day of the week, a type of whether it is a public holiday, and a type of whether it is the end of the month, in the prediction target period; acquiring production schedule information of a target consumer that produces a product for the product during the forecast target period; a prediction step of predicting the amount of electricity demand of the target consumer during the prediction period by inputting the first calendar information and the production schedule information into a machine learning model generated by performing machine learning using second calendar information indicating a date and time when electricity was consumed in at least one consumer, the second calendar information including at least one of the day of the week, whether it is a public holiday, and whether it is the end of the month at the date and time, and production performance information of the at least one consumer at the date and time as explanatory variables, and the amount of electricity consumed by the at least one consumer at the date and time as a target variable; A power demand forecasting method comprising: a first power load, a second power load, and a control device that controls the first power load are installed in the at least one consumer; The objective variable is the power consumption amount obtained by subtracting the power consumption amount of the first power load from the sum of the power consumption amount of the first power load and the power consumption amount of the second power load at the date and time. Electricity demand forecasting methods.

6. a first acquisition means for acquiring first calendar information indicating a prediction target period and including at least one of a day of the week, a type indicating whether it is a public holiday, and a type indicating whether it is the end of the month, in the prediction target period; a second acquisition means for acquiring production schedule information of a target consumer that produces a product during the forecast target period; a prediction means for predicting the amount of electricity demand of the target consumer during the prediction period by inputting the first calendar information and the production schedule information into a machine learning model generated by performing machine learning using second calendar information indicating the date and time when electricity was consumed in at least one consumer, the second calendar information including at least one of the day of the week, whether it is a public holiday, and whether it is the end of the month at the date and time, and production performance information of the at least one consumer at the date and time as explanatory variables, and the amount of electricity consumed by the at least one consumer at the date and time as a target variable; An information processing device comprising: a first power load, a second power load, and a control device that controls the first power load are installed in the at least one consumer; The objective variable is the power consumption amount obtained by subtracting the power consumption amount of the first power load from the sum of the power consumption amount of the first power load and the power consumption amount of the second power load at the date and time. Information processing device.

Images