JP6476338B1 - Power demand forecasting system, power demand forecasting model construction method, program, and sales support system - Google Patents

Abstract

A power demand prediction system useful for predicting power demand of a consumer whose power usage record is unknown is provided. A power demand prediction system includes a learning database for storing existing customer data in which consumer attribute data and identification data of a power demand pattern to which a customer's power use record belongs are associated, and a consumer. A model building unit that builds a power demand prediction model that outputs pattern prediction data indicating which of the plurality of power demand patterns uses power according to the input of the attribute data of the consumer based on the learning database 124 125, a prediction data acquisition unit 224 that inputs the attribute data of the new consumer into the power demand prediction model and acquires the pattern prediction data of the consumer, and the new consumer has a plurality of electric powers based on the pattern prediction data. A prediction result output unit 225 that outputs a prediction result as to which of the demand patterns uses power. [Selection] Figure 2

Description

  The present disclosure relates to a power demand prediction system, a power demand prediction model construction method, a program, and a sales support system.

  Patent Document 1 discloses a means for calculating consumer feature quantities based on attribute information indicating consumer attributes, a means for classifying consumers into clusters based on consumer feature quantities, and a consumer used in the past. An apparatus is disclosed that includes means for calculating the average amount of power used for each cluster, and means for providing information indicating the average power consumption calculated for each cluster to consumers.

JP 2017-120519 A

  The present disclosure provides a power demand prediction system that is useful for predicting the power demand of a consumer whose power usage record is unknown.

  A power demand prediction system according to an aspect of the present disclosure includes a pattern determination unit that determines which of a plurality of preset power demand patterns a power usage record by a consumer belongs to, and a plurality of types of data items A database that stores existing customer data that associates the attribute data of the consumer with the identification data of the power demand pattern to which the power usage record of the consumer belongs, and the consumer uses any of the power demand patterns A model building unit that builds a power demand prediction model that outputs pattern prediction data that indicates whether to use the data in response to the input of the attribute data of the consumer based on the database, and the attribute data of the new consumer Based on the prediction data acquisition unit that inputs into the prediction model and acquires the pattern prediction data of the consumer, and the pattern prediction data of the new consumer There are provided a prediction result output unit for outputting one of the prediction result the new customer to use power in any of a plurality of power demand pattern.

  According to this power demand forecasting system, a highly reliable power demand forecasting model is created by constructing a power demand forecasting model after classifying and simplifying the data used by consumers into a plurality of power demand patterns. Can be easily constructed. And by inputting the attribute data of a new consumer into the power demand prediction model, it is possible to easily predict which of the plurality of power demand patterns will use the power. Therefore, this power demand prediction system is useful for predicting the power demand of a consumer whose power usage record is unknown.

  The model construction unit may construct a power demand prediction model by machine learning based on a database. In this case, a more reliable power demand prediction model can be constructed by machine learning.

  Each of the plurality of power demand patterns may be determined by a combination of at least a transition pattern of the power usage amount in one day and a transition pattern of the power usage amount in one year. In this case, it is possible to predict both the trend of power consumption in one day and the trend of power consumption in one year based on the prediction result of which of the multiple power demand patterns a consumer uses. Therefore, it is possible to accurately predict the transition of the power consumption of the consumer.

  The pattern determination unit further determines to which of a plurality of preset power demand patterns the power usage record by the new customer belongs, and the database includes the attribute data of the new customer, the pattern determination unit In addition, the existing customer data for update that correlates with the power demand pattern of the relevant consumer determined by (1) is further accumulated, and the model building unit uses the machine learning based on the database that stores the existing customer data for update. The demand prediction model may be updated. In this case, since the power demand prediction model is continuously brushed up based on the actual power usage by new consumers, the reliability of the prediction result of the power demand pattern based on the power demand prediction model is improved.

  The power demand prediction system may further include a contribution calculation unit that calculates a contribution indicating the degree of influence on the pattern prediction data for each of a plurality of types of data items based on the power demand prediction model. In this case, based on the magnitude of the contribution degree, it is possible to narrow down data items to be input to the power demand prediction model while suppressing a decrease in reliability of the prediction result of the power demand pattern. Thereby, the data item which should be acquired from a new consumer can be reduced, and the prediction of the power demand pattern about the said consumer can be made still easier.

  The method for constructing a power demand prediction model according to another aspect of the present disclosure includes determining whether a power usage record by a consumer belongs to a plurality of preset power demand patterns, and a plurality of types of data items Existing customer data in which the attribute data of the consumer including the data and the identification data of the power demand pattern to which the power usage record of the consumer belongs are associated with each other in the database. And building a power demand prediction model that outputs pattern prediction data indicating which power is used according to the input of the attribute data of the consumer based on the database.

  A program according to still another aspect of the present disclosure includes a customer demand data including a plurality of types of data items and a power demand pattern to which a power usage record by the consumer belongs among a plurality of preset power demand patterns. Pattern prediction data indicating which of the plurality of power demand patterns the power is used by the consumer based on the database that stores the existing customer data in association with the identification data of Input new customer attribute data into a power demand prediction model that is constructed to output in response to input to obtain pattern prediction data for the customer, and Based on this, the apparatus executes to output a prediction result as to which of the plurality of power demand patterns the new consumer uses power.

  A sales support system according to still another aspect of the present disclosure includes a pattern determination unit that determines which of a plurality of preset power demand patterns a power usage record by a consumer, and a plurality of types of data items. A database that stores existing customer data that associates the attribute data of the consumer including the identification data of the power demand pattern to which the power usage record of the consumer belongs, and the consumer is any of the plurality of power demand patterns A model building unit that builds a power demand prediction model that outputs pattern prediction data indicating whether to use power in accordance with the input of the attribute data of the consumer based on the database, and attribute data of the new consumer A prediction data acquisition unit that inputs into the power demand prediction model and acquires pattern prediction data of the consumer, and a pattern prediction data of a new consumer And a plan output section for outputting a power subscription plan suitable power demand pattern shown by. This system is useful for proposing a reasonable power usage contract plan for a consumer whose power usage is unknown.

  According to the present disclosure, it is possible to provide a power demand prediction system that is useful for predicting the power demand of a consumer whose power usage record is unknown.

It is a schematic diagram which illustrates schematic structure of an electric power demand prediction system. It is a block diagram which illustrates the functional structure of an electric power demand prediction system. It is a schematic diagram which illustrates an electric power demand pattern. It is a schematic diagram which illustrates notionally the electric power demand prediction model constructed | assembled by the random forest method. It is a flowchart which illustrates the setting procedure of an electric power demand pattern. It is a flowchart which illustrates the construction procedure of an electric power demand prediction model. It is a flowchart which illustrates the output procedure of an electric power demand prediction result.

  Hereinafter, embodiments will be described in detail with reference to the drawings. In the description, the same elements or elements having the same functions are denoted by the same reference numerals, and redundant description is omitted.

[Power demand forecasting system]
The power demand prediction system 1 according to the present embodiment is a system that supports prediction of power demand of a consumer whose power usage record is unknown. As shown in FIG. 1, the power demand prediction system 1 includes a prediction server 100 and a prediction terminal 200.

  The prediction server 100 constructs a power demand prediction model for predicting which of the plurality of power demand patterns set in advance by the consumer based on the input of consumer attribute data. . For example, the prediction server 100 constructs a power demand prediction model based on the attribute data of the consumer 10 and the power usage record data of the consumer 10. The prediction server 100 acquires the power usage record data of the customer 10 from the use record server 20.

  The usage record server 20 is connected to the watt hour meters 11 of a plurality of consumers 10 via a network NW such as the Internet, and is a server that acquires and accumulates power usage record data from each watt hour meter 11. is there. Note that the prediction server 100 may acquire power usage record data from the watt hour meter 11 of the consumer 10 without using the usage record server 20.

  For example, the prediction server 100 includes a circuit 110. The circuit 110 includes a processor 111, a memory 112, a storage 113, and a communication port 114. The storage 113 is a non-volatile storage medium (for example, a hard disk or a flash memory) that can be read by a computer. The storage 113 determines which of a plurality of preset power demand patterns the power usage record by the customer 10 belongs to, attribute data of the customer 10 including a plurality of types of data items, and the demand In response to accumulating in the database existing customer data in which the identification data of the power demand pattern to which the power usage record of the house 10 belongs is associated with the input of the attribute data of the consumer 10 A program for causing a device to execute a power demand prediction model that outputs pattern prediction data indicating which power demand pattern is used by machine learning based on a database is stored.

  The memory 112 temporarily stores a program loaded from the storage 113, a calculation result by the processor 111, and the like. The processor 111 executes the program in cooperation with the memory 112. The processor 111 may be configured by a plurality of circuit elements. The communication port 114 is connected to the usage record server 20 and the prediction terminal 200 via the network NW, and communicates information between the usage record server 20 and the prediction terminal 200 according to a command from the processor 111. I do. The connection here means that information communication is possible, and does not necessarily mean a physical connection by wire. The same applies to the following. That is, at least a part of the network NW may be a wireless communication path.

  The prediction terminal 200 is a terminal for using the power demand prediction model constructed by the prediction server 100. For example, the prediction terminal 200 includes a main body 201, a display device 202, and an input device 203. The display device 202 is a device for displaying information for the user. Specific examples of the display device 202 include a liquid crystal monitor and an organic EL monitor. The input device 203 is a device for acquiring input by the user. Specific examples of the input device 203 include a keyboard, a touch pad, and a mouse. Note that the display device 202 and the input device 203 may be integrated as a so-called touch panel. Further, the display device 202 and the input device 203 may be integrated with the main body 201.

  The main body 201 includes a circuit 210. The circuit 210 includes a processor 211, a memory 212, a storage 213, a communication port 214, and an input / output port 215. The storage 213 is a computer-readable non-volatile storage medium (for example, a hard disk or a flash memory). The storage 213 inputs new consumer attribute data to the power demand prediction model to acquire the pattern prediction data of the consumer, and the new demand based on the pattern prediction data of the new consumer. A program for causing the apparatus to output (for example, display on the display device 202) a prediction result as to which of the plurality of power demand patterns is used by the house is stored. The communication port 214 is connected to the prediction server 100 or the like via the network NW, and performs information communication with the prediction server 100 or the like in response to a command from the processor 211. The input / output port 215 inputs / outputs information between the display device 202 and the input device 203 in accordance with a command from the processor 211.

  Hereinafter, the configuration of the prediction server 100 and the prediction terminal 200 will be described in more detail. FIG. 2 is a block diagram illustrating a functional configuration of the prediction server 100 and the prediction terminal 200. As shown in FIG. 2, the prediction server 100 includes a performance classification unit 131, a classification database 132, a pattern holding unit 121, and a pattern determination unit as functional configurations (hereinafter referred to as “functional modules”). 122, a data registration unit 123, a learning database 124, a model construction unit 125, a model holding unit 126, a contribution calculation unit 127, and a prediction data generation unit 128.

  The result classifying unit 131 acquires data on the power use results of the consumer from the use result server 20 and accumulates the data in the classification database 132, and sets a plurality of power demand patterns by the classification process of the data accumulated in the classification database 132. And registered in the pattern holding unit 121. For example, the performance classifying unit 131 sets a plurality of power demand patterns determined by a combination of transition data of power usage in one day and transition data of power usage in one year.

  For example, the performance classifying unit 131 classifies the transition data (hereinafter referred to as “day characteristic data”) of the power consumption in one day into a plurality of daily characteristic groups, and the transition data (hereinafter “ "Monthly characteristic data") is classified into a plurality of monthly characteristic groups, and a plurality of power demand patterns determined by combinations of a plurality of daily characteristic groups and a plurality of monthly characteristic groups are set.

  For example, the daily characteristic data is composed of 48 data indicating the amount of power used every 30 minutes, and each data is dimensionless with the maximum value of 48 data being 1. The performance classifying unit 131 classifies the past day characteristic data of a plurality of consumers into a plurality of (for example, three) day characteristic groups L1-1, L1-2, and L1-3 by a classification method such as a K-average method. (See FIG. 3). The specific numerical value to be classified by the K average method or the like may be the total value of the 48 data or the average value of the 48 data. As a result of the classification by the K-average method or the like, in the group L1-2 in FIG. 3, the ratio of the night load (daytime power usage) to the night load (night power usage) is higher than the group L1-1. ing. In the group L1-3, the ratio of the night load to the day load is higher than that of the group L1-2.

  For example, the monthly characteristic data is composed of 12 data indicating the amount of power used every month, and each data is made dimensionless with the maximum value of 12 data being 1. The performance classifying unit 131 classifies past month characteristic data of a plurality of customers into a plurality of (for example, three) month characteristic groups L2-1, L2-2, and L2-3 by a classification method such as a K-average method. (See FIG. 3). The specific numerical value to be classified by the K-average method or the like may be the total value of the 12 data or the average value of the 12 data. As a result of the classification by the K-average method or the like, in the group L2-1 in FIG. 3, the peak of power consumption tends to be concentrated at one point (for example, August). In the group L2-2, there is a tendency that the peak of power usage is concentrated at two points (for example, February and August). In the group L2-3, there is a tendency that the fluctuation of the power consumption is small throughout the year.

  The pattern determination unit 122 determines which of the plurality of power demand patterns stored in the pattern holding unit 121 (a plurality of power demand patterns preset by the result classification unit 131) stored by the consumer. . The data registration unit 123 associates the attribute data of the consumer with the power demand pattern to which the power usage record of the consumer belongs (the power demand pattern of the consumer determined by the pattern determination unit 122) (hereinafter, referred to as “data demand pattern”). "Existing customer data") is registered in the learning database 124. The attribute data includes a plurality of data items that define the nature of the consumer. The plurality of data items include, for example, the presence / absence of a store in the customer's building, the presence / absence of a 24-hour store (eg, a convenience store) in the customer's building, the floor area of the customer's building, and the floor area of the customer's building. , The number of floors of the customer's building, the air conditioning system of the customer's building (for example, central air conditioning or individual air conditioning, etc.), and the age of the building of the customer.

  The learning database 124 (database) stores the existing customer data registered by the data registration unit 123. The model construction unit 125 is configured to learn a power demand prediction model that outputs pattern prediction data indicating which of the plurality of power demand patterns the electric power is used by the consumer according to the input of consumer attribute data. Build based on database 124. For example, the model construction unit 125 constructs a power demand prediction model by machine learning using the existing customer data accumulated in the learning database 124 as teacher data. Machine learning here means an information processing method that builds a power demand prediction model by repeatedly adjusting the parameters of the power demand prediction model so that the input and output of the power demand prediction model match the customer data. To do. Specific examples of such an information processing method include a known method such as a random forest method and deep learning.

  FIG. 4 is a schematic diagram conceptually showing a power demand prediction model constructed by the random forest method. The model construction unit 125 randomly generates a plurality of decision trees T1 to Tn based on the attribute data included in the teacher data, and constructs each of the decision trees T1 to Tn based on the teacher data. For example, the model construction unit 125 constructs a decision tree to be constructed (hereinafter referred to as “decision tree T”) as follows. That is, the model construction unit 125 randomly samples the teacher data, gives the sampled data to the highest node N1 of the decision tree T, and obtains a branch function of the node N1. Next, based on the branch function, the teacher data is divided into two systems and given to the two nodes N2 having the node N1 as a parent, and the branch functions of the two nodes N2 are obtained. Thereafter, the above processing is continued in the lower nodes until the preset termination condition is satisfied, and the construction of the decision tree T is completed.

  Returning to FIG. 2, the model holding unit 126 stores data of the power demand prediction model constructed by the model construction unit 125. The contribution calculation unit 127 calculates a contribution indicating the degree of influence on the pattern prediction data for each of a plurality of types of data items based on the power demand prediction model stored in the model holding unit 126. As the contribution of the data item increases, the possibility that the pattern prediction data differs depending on the difference in the value of the data item increases. The degree of contribution for each of the plurality of types of data items can be derived based on, for example, the structure of the decision tree in the power demand prediction model and the content of the branch function of each node.

  The prediction data generation unit 128 inputs the consumer attribute data to the power demand prediction model stored in the model holding unit 126 to generate the pattern prediction data of the consumer. For example, the prediction data generation unit 128 randomly extracts a plurality of decision trees from the power demand prediction model, and among the plurality of types of outputs obtained by inputting attribute data to each of the extracted plurality of decision trees, The output from the decision tree is generated as pattern prediction data.

  Note that the prediction server 100 may be configured to update the power demand prediction model based on data acquired after the construction of the power demand prediction model. For example, the pattern determination unit 122 further determines to which of the plurality of power demand patterns the power usage record by a new consumer (a consumer added after the construction of the power demand prediction model) belongs, and the learning database 124. Further accumulates the existing customer data for update in which the attribute data of the new customer is associated with the power demand pattern of the customer determined by the pattern determination unit 122, and the model construction unit 125 The power demand prediction model is updated by machine learning based on the learning database 124 that stores the existing customer data.

  Each functional module of the prediction server 100 is configured by, for example, the processor 111 executing a program in the storage 113 in cooperation with the memory 112. Storage units such as the pattern holding unit 121, the learning database 124, and the model holding unit 126 are configured in a partial storage area of the storage 113. The prediction server 100 is not necessarily limited to one in which each functional module is configured by a program. At least some of the functional modules of the prediction server 100 may be configured by a dedicated logic circuit or an ASIC (Application Specific Integrated Circuit) in which the logic circuit is integrated.

  The prediction terminal 200 includes a contribution data acquisition unit 221, an interface image display unit 222, an attribute data acquisition unit 223, a prediction data acquisition unit 224, and a prediction result output unit 225 as functional modules.

  The contribution degree data acquisition unit 221 acquires contribution degree data calculated by the contribution degree calculation unit 127. The interface image display unit 222 displays an interface image for using the power demand prediction model (for example, displayed on the display device 202). The interface image display unit 222 includes at least an input unit for inputting customer identification information and a plurality of data items of customer attribute data, and a tool (for example, a button) for requesting generation of pattern prediction data. An interface image including an image is displayed. The interface image display unit 222 may display the interface image so that the contribution information acquired by the contribution data acquisition unit 221 can be recognized. For example, in the interface image, the interface image display unit 222 may display the contribution degree for each of the plurality of data items as a numerical value or the like, or may display the input unit by narrowing down to data items having a higher contribution degree. Good.

  The attribute data acquisition unit 223 acquires the attribute data input to the interface image display unit 222. The prediction data acquisition unit 224 inputs the consumer attribute data acquired by the attribute data acquisition unit 223 to the power demand prediction model and acquires the pattern prediction data of the consumer. For example, the prediction data acquisition unit 224 transmits the consumer attribute data acquired by the attribute data acquisition unit 223 to the prediction data generation unit 128 and inputs the attribute data to the demand prediction model to predict the pattern prediction data generated. Received from the data generation unit 128.

  Based on the consumer pattern prediction data acquired by the prediction data acquisition unit 224, the prediction result output unit 225 outputs a prediction result indicating which of the plurality of power demand patterns the power uses (for example, the above) Display on the interface image).

  Each functional module of the prediction terminal 200 is configured, for example, by the processor 211 executing a program in the storage 213 in cooperation with the memory 212. The prediction terminal 200 is not necessarily limited to one in which each functional module is configured by a program. At least some of the functional modules of the prediction terminal 200 may be configured by a dedicated logic circuit or an ASIC in which this is integrated.

[Power demand forecast procedure]
Then, the electric power demand prediction procedure which the electric power demand prediction system 1 performs is illustrated. This procedure includes a procedure for setting a plurality of power demand patterns executed by the prediction server 100, a procedure for building a power demand prediction model (a method for building a power demand prediction model), and a power demand prediction result executed by the prediction terminal 200. Output procedure. Hereinafter, these procedures are illustrated in detail.

(Pattern setting procedure)
The setting procedure of the power demand pattern executed by the prediction server 100 includes classifying the data on the power usage record of the consumer (hereinafter referred to as “usage record data”) into a plurality of power demand patterns. As shown in FIG. 5, the prediction server 100 first executes steps S21 and S22. In step S <b> 21, the result classification unit 131 uses the use record data (for example, data of the power use record for one year) of any one of the consumers (hereinafter referred to as “data accumulation target consumer”). Get from. In step S <b> 22, the result classifying unit 131 confirms whether or not the data of the customer to be stored is already stored in the classification database 132.

  In step S22, when it is determined that the usage record data of the data accumulation target consumer is already accumulated in the classification database 132, the prediction server 100 executes step S23. In step S23, the result classification unit 131 overwrites the registered use record data with the use record data acquired in step S21.

  In Step S22, when it is determined that the usage record data of the data storage target consumer is not stored in the classification database 132, the prediction server 100 executes Step S24. In step S <b> 24, the result classification unit 131 newly registers the use result data of the data accumulation target consumer in the classification database 132. When the set power demand pattern is already stored in the pattern holding unit 121, the usage record data registered by the track record classification unit 131 in step S24 corresponds to “updated usage track record data” described later.

  Following step S23 or step S24, the prediction server 100 executes step S25. In step S <b> 25, the result classification unit 131 confirms whether or not the use result data has been acquired for all the customers who can acquire the use result data.

  If it is determined in step S25 that acquisition of usage record data has not been completed for all consumers, the prediction server 100 returns the process to step S21. Thereafter, until the acquisition of the usage record data for all the consumers is completed, the update or new registration of the use record data is repeated while changing the data accumulation target consumer.

  If it is determined in step S25 that acquisition of usage record data has been completed for all consumers, the prediction server 100 executes step S26. In step S26, the model construction unit 125 determines whether the number of usage record data accumulated in the learning database 124 has reached the number of classification data (for example, the number of data suitable for setting a predetermined number of power demand patterns). Confirm whether or not. When the set power demand pattern is stored in the pattern holding unit 121, the number of update usage record data accumulated after setting the power demand pattern is the number of data for classification (for example, for updating the power demand pattern). Check if the appropriate number of data has been reached.

  If it is determined in step S26 that the number of usage record data has reached the number of classification data, the prediction server 100 executes step S27. In step S <b> 27, the performance classification unit 131 sets a plurality of power demand patterns by the classification process of the data stored in the classification database 132 (for example, the classification process by the K average method), and registers it in the pattern holding unit 121. When the set power demand pattern is stored in the pattern holding unit 121, the result classifying unit 131 updates a plurality of power demand patterns by the classification process including the update use result data. If the number of usage record data does not reach the number of classification data in step S26, the prediction server 100 completes the power demand pattern setting procedure without executing step S27. The prediction server 100 repeats the above procedure at a predetermined cycle.

(Procedure for building a power demand prediction model)
The construction procedure of the power demand prediction model executed by the prediction server 100 is to determine which of the plurality of preset power demand patterns the power usage record by the consumer belongs to, and a plurality of types of data items. The existing customer data in which the attribute data of the consumer including the data and the identification data of the power demand pattern to which the power usage record of the consumer belongs is associated with the database, and according to the input of the customer attribute data And constructing, by machine learning based on a database, a power demand prediction model that outputs pattern prediction data indicating which of the plurality of power demand patterns is used by the consumer.

  As shown in FIG. 6, the prediction server 100 first executes steps S01, S02, and S03. In step S01, the pattern determining unit 122 uses the power usage record data of one of the consumers (hereinafter referred to as “data storage target consumer”) (for example, the power usage record data for one year). Obtain from the server 20. In step S02, the pattern determination unit 122 determines which of the plurality of power demand patterns the power usage record by the data storage target consumer belongs to based on the data acquired in step S01. In step S <b> 03, the data registration unit 123 confirms whether or not the data of the data accumulation target customer has already been accumulated in the learning database 124.

  In Step S03, when it is determined that the data of the consumer for which data is to be accumulated has already been accumulated in the learning database 124, the prediction server 100 executes Step S04. In step S04, the data registration part 123 overwrites the identification data of the power demand pattern determined in step S02 on the already registered existing customer data of the data storage target consumer.

  In Step S03, when it is determined that the data of the customer for which data is to be accumulated is not accumulated in the learning database 124, the prediction server 100 executes Step S05. In step S05, the data registration unit 123 newly adds existing customer data in the learning database 124 in which the attribute data of the data storage target consumer and the power demand pattern identification data determined in step S02 are associated with each other. sign up. In addition, when the constructed power demand prediction model is already stored in the model holding unit 126, the existing customer data registered by the data registration unit 123 in step S05 is the above-mentioned “existing customer data for update”. Equivalent to.

  Following step S04 or step S05, the prediction server 100 executes step S06. In step S <b> 06, the pattern determination unit 122 confirms whether or not the acquisition of power usage data has been completed for all customers who can acquire power usage data.

  If it is determined in step S06 that acquisition of power usage data has not been completed for all consumers, the prediction server 100 returns the process to step S01. Thereafter, until acquisition of power usage data for all the consumers is completed, the update or new registration of existing customer data is repeated while changing the data accumulation target consumers.

  If it is determined in step S06 that acquisition of power usage data has been completed for all consumers, the prediction server 100 executes step S07. In step S07, the model construction unit 125 confirms whether or not the number of existing customer data accumulated in the learning database 124 has reached the number of learning data (for example, the number of data suitable for model construction). To do. When the built power demand prediction model is stored in the model holding unit 126, the number of existing customer data for update accumulated after the model is built is the number of data for learning (for example, suitable for model update). Check whether the number of data) has been reached.

  When it is determined in step S07 that the number of existing customer data has reached the number of learning data, the prediction server 100 executes steps S08 and S09. In step S08, the model construction unit 125 constructs a power demand prediction model by machine learning based on the learning database 124 and registers it in the model holding unit 126. When the constructed power demand prediction model is stored in the model holding unit 126, the model construction unit 125 updates the power demand prediction model by machine learning based on the learning database 124 that stores the existing customer data for update. To do. In step S09, the contribution calculation unit 127 calculates a contribution indicating the degree of influence on the pattern prediction data for each of a plurality of types of data items based on the power demand prediction model stored in the model holding unit 126. This completes the procedure for constructing a power demand prediction model. If the number of existing customer data does not reach the number of learning data in step S07, the prediction server 100 completes the power demand prediction model construction procedure without executing steps S08 and S09. The prediction server 100 repeats the above procedure at a predetermined cycle.

(Procedure for outputting power demand prediction results)
The output procedure of the power demand prediction result executed by the prediction terminal 200 includes inputting new customer attribute data to the power demand prediction model to acquire the pattern prediction data of the customer, and creating a new demand. Outputting a prediction result as to which of the plurality of power demand patterns the new consumer uses power based on the house pattern prediction data.

  As shown in FIG. 7, the prediction terminal 200 first executes steps S11, S12, S13, and S14. In step S <b> 11, the contribution data acquisition unit 221 requests the contribution calculation unit 127 to transmit contribution data. In step S12, the contribution degree data acquisition unit 221 acquires the contribution degree data transmitted by the contribution degree calculation unit 127. In step S <b> 13, the interface image display unit 222 displays the interface image on the display device 202. In step S14, the attribute data acquisition unit 223 waits for command input of pattern prediction data. For example, the attribute data acquisition unit 223 waits for a tool for requesting generation of pattern prediction data to be operated on the interface image.

  Next, the prediction terminal 200 executes steps S15, S16, and S17. In step S <b> 15, the attribute data acquisition unit 223 acquires the attribute data input to the interface image display unit 222. In step S16, the prediction data acquisition unit 224 inputs the consumer attribute data acquired by the attribute data acquisition unit 223 into the power demand prediction model, and acquires the pattern prediction data of the consumer. For example, the prediction data acquisition unit 224 transmits the consumer attribute data acquired by the attribute data acquisition unit 223 to the prediction data generation unit 128 and inputs the attribute data to the demand prediction model to predict the pattern prediction data generated. Received from the data generation unit 128. In step S <b> 17, based on the consumer pattern prediction data acquired by the prediction data acquisition unit 224, the prediction result output unit 225 displays a prediction result as to which of the plurality of power demand patterns the customer uses power. Output (for example, display on the interface image). Thus, the output procedure of the power demand prediction result is completed.

[Effect of this embodiment]
As described above, the power demand prediction system 1 includes the pattern determination unit 122 that determines which of the plurality of preset power demand patterns the power usage record by the customer 10 belongs to, and a plurality of types of data. A database for learning 124 that stores existing customer data in which attribute data of the customer 10 including items and identification data of a power demand pattern to which the power usage record of the customer 10 belongs are associated; In response to input of attribute data, a power demand prediction model that outputs pattern prediction data indicating which of the plurality of power demand patterns is used by the customer 10 is constructed by machine learning based on the learning database 124 Model building unit 125 that inputs the attribute data of the new customer 10 to the power demand prediction model, and Based on the prediction data acquisition unit 224 that acquires the current prediction data and the pattern prediction data of the new consumer 10, the prediction result of which of the plurality of power demand patterns the new consumer 10 uses power A prediction result output unit 225.

  According to the power demand forecasting system 1, by using machine learning after classifying and simplifying the data of actual usage by the customer 10 into a plurality of power demand patterns, a highly reliable power demand forecasting model can be facilitated. Can be built. And by inputting the attribute data of the new consumer 10 to a power demand prediction model, it can be easily predicted in which of the plurality of power demand patterns the consumer 10 uses power. Therefore, the power demand prediction system 1 is useful for predicting the power demand of the customer 10 whose power usage record is unknown.

  Each of the plurality of power demand patterns may be determined by a combination of at least a transition pattern of the power usage amount in one day and a transition pattern of the power usage amount in one year. In this case, based on the prediction result of which of the plurality of power demand patterns the customer 10 uses the power, both the transition of the power usage in the day and the transition of the power usage in the year are shown. Since it becomes possible to predict, the transition of the power consumption of the consumer 10 can be accurately predicted.

  The pattern determination unit 122 further determines to which of a plurality of preset power demand patterns the power usage record by the new consumer 10 belongs, and the learning database 124 stores the attribute data of the new consumer. And the existing customer data for updating in association with the power demand pattern of the customer 10 determined by the pattern determining unit 122, and the model building unit 125 stores the existing customer data for updating. The power demand prediction model may be updated by machine learning based on the learned database 124. In this case, since the power demand prediction model is continuously brushed up based on the actual power usage by the new customer 10, the reliability of the prediction result of the power demand pattern based on the power demand prediction model is improved.

  The power demand prediction system 1 may further include a contribution calculation unit 127 that calculates a contribution indicating the degree of influence on the pattern prediction data for each of a plurality of types of data items based on the power demand prediction model. In this case, based on the magnitude of the contribution degree, it is possible to narrow down data items to be input to the power demand prediction model while suppressing a decrease in reliability of the prediction result of the power demand pattern. Thereby, the data item which should be acquired from the new consumer 10 can be reduced, and the prediction of the power demand pattern about the said consumer 10 can be made still easier.

  Although the embodiment has been described above, the present invention is not necessarily limited to the above-described embodiment, and various modifications can be made without departing from the scope of the invention.

  DESCRIPTION OF SYMBOLS 1 ... Electric power demand prediction system, 10 ... Consumer, 122 ... Pattern determination part, 124 ... Database for learning (database), 125 ... Model construction part, 127 ... Contribution calculation part, 224 ... Prediction data acquisition part, 225 ... Prediction Result output part.

Claims (8)

A pattern determination unit that determines which of a plurality of preset power demand patterns the power usage record by a consumer belongs to;
A database for accumulating existing customer data in which the attribute data of the consumer including a plurality of types of data items and the identification data of the power demand pattern to which the power usage record of the consumer belongs are associated with each other;
Based on the database, a power demand prediction model that outputs pattern prediction data indicating which of the plurality of power demand patterns the electric power is used by the consumer according to the input of the attribute data of the consumer A model building department to
A prediction data acquisition unit that inputs new consumer attribute data into the power demand prediction model and acquires the pattern prediction data of the consumer;
A power demand prediction comprising: a prediction result output unit that outputs a prediction result as to which of the plurality of power demand patterns the new consumer uses power based on the pattern prediction data of the new consumer system.   The power demand prediction system according to claim 1, wherein the model construction unit constructs the power demand prediction model by machine learning based on the database.   The power demand according to claim 1 or 2, wherein each of the plurality of power demand patterns is defined by a combination of at least a transition pattern of power usage in a day and a transition pattern of power usage in a year. Prediction system. The pattern determination unit further determines to which of the plurality of preset power demand patterns the power usage record by a new consumer belongs,
The database further accumulates existing customer data for update in which the attribute data of the new customer and the power demand pattern of the customer determined by the pattern determination unit are associated with each other,
The power demand prediction system according to any one of claims 1 to 3, wherein the model construction unit updates the power demand prediction model by machine learning based on the database in which existing customer data for update is accumulated.   The contribution calculation part which calculates the contribution which shows the grade of the influence on the said pattern prediction data for every several types of data item based on the said electric power demand prediction model is further provided. The power demand forecasting system described. Determining which of the plurality of preset power demand patterns the power usage record by the consumer belongs to;
Storing in the database existing customer data in which attribute data of the consumer including a plurality of types of data items and identification data of the power demand pattern to which the customer's power usage record belongs are associated with each other;
Based on the database, a power demand prediction model that outputs pattern prediction data indicating which of the plurality of power demand patterns the electric power is used by the consumer according to the input of the attribute data of the consumer And a method for constructing a power demand prediction model. Existing demand in which consumer attribute data including a plurality of types of data items is associated with identification data of the power demand pattern to which a power usage record by the consumer belongs among a plurality of preset power demand patterns Based on a database in which house data is accumulated, pattern prediction data indicating which of the plurality of power demand patterns is used by the consumer according to the input of the attribute data of the consumer is output. Inputting the attribute data of a new customer into the constructed power demand prediction model to obtain the pattern prediction data of the customer;
A program that causes the apparatus to execute a prediction result indicating which of the plurality of power demand patterns the new consumer uses power based on the pattern prediction data of the new consumer. A pattern determination unit that determines which of a plurality of preset power demand patterns the power usage record by a consumer belongs to;
A database for accumulating existing customer data in which the attribute data of the consumer including a plurality of types of data items and the identification data of the power demand pattern to which the power usage record of the consumer belongs are associated with each other;
Based on the database, a power demand prediction model that outputs pattern prediction data indicating which of the plurality of power demand patterns the electric power is used by the consumer according to the input of the attribute data of the consumer A model building department to
A prediction data acquisition unit that inputs new consumer attribute data into the power demand prediction model and acquires the pattern prediction data of the consumer;
A sales support system comprising: a plan output unit that outputs a power usage contract plan suitable for a power demand pattern indicated by the pattern prediction data of a new consumer.

Images