JP6908444B2 - Power Demand Forecasting Device, Power Demand Forecasting Method and Program - Google Patents

Description

The present invention relates to a power demand forecasting device, a power demand forecasting method and a program.

In the retail electricity business, the demand forecasting function is an important function that influences the income and expenditure of the business, and improvement of the demand forecasting function is required. Therefore, various demand forecasting systems and demand forecasting methods have been proposed (see, for example, Patent Document 1).

Japanese Unexamined Patent Publication No. 2013-066318

The power demand forecast system described in Patent Document 1 discloses a technique of reading information such as a calendar stored in a recording medium or the like and constructing a demand forecast model based on the read information or the like.

By using the calendar information as described above, it is possible to predict changes in the power demand of the facility depending on the day of the week and public holidays.
However, especially in facilities where regular holidays are different from general holidays (Saturdays, Sundays, national holidays, etc.), there is a problem that the accuracy of forecasting power demand is lowered if the calendar information is not set appropriately. That is, if the calendar information contains an error or the calendar information is not updated properly, there is a problem that the prediction accuracy of the power demand based on the calendar information is lowered.

The present invention has been made to solve the above problems, and an object of the present invention is to provide a power demand forecasting device, a power demand forecasting method, and a program that facilitates ensuring the prediction accuracy of power demand.

In order to achieve the above object, the present invention provides the following means.
The electric power demand forecasting device according to the first aspect of the present invention includes an acquisition unit that acquires a plurality of demand actual information which is information on electric power demand measured in a predetermined period, and a plurality of acquired electric power demand forecasting devices based on the acquired plurality of the acquired actual demand information. The analysis unit for obtaining the demand pattern of the above, the division unit for allocating the obtained plurality of demand patterns to a plurality of predetermined divisions, and each of the plurality of demand actual information can be assigned to any of the plurality of demand patterns. Corresponds to the comparison unit for determining whether or not to correspond, the selection unit for selecting the category assigned to the demand pattern determined to correspond to each of the plurality of demand record information, and the demand record information. For the predetermined period to be performed, the selected division and the actual table setting unit for setting the actual table to which the division pattern which is the information of the electric power demand associated with the division is assigned are provided. It is characterized by being.

The electric power demand forecasting method according to the second aspect of the present invention includes a plurality of acquisition steps for acquiring a plurality of demand actual information which is information on electric power demand measured in a predetermined period, and a plurality of acquired power demand forecast methods based on the acquired plurality of the acquired actual demand information. An analysis step for obtaining the demand pattern of the above, a division step for allocating the obtained plurality of demand patterns to a plurality of predetermined categories, and each of the plurality of demand actual information can be assigned to any of the plurality of demand patterns. Corresponds to the comparison step for determining whether to correspond, the selection step for selecting the category assigned to the demand pattern determined to correspond in the comparison step for each of the plurality of demand actual information, and the demand actual information. To have the selected division and the actual table setting step for setting the actual table to which the division pattern which is the information of the electric power demand associated with the division is assigned for the predetermined period. It is a feature.

A plurality of programs according to a third aspect of the present invention include an acquisition function for acquiring a plurality of demand record information which is information on electric power demand measured in a predetermined period, and a plurality of programs based on the acquired plurality of the acquired demand record information. The analysis function for obtaining the demand pattern of the above, the division function for allocating the obtained plurality of demand patterns to a plurality of predetermined categories, and each of the plurality of demand actual information can be assigned to any of the plurality of demand patterns. Corresponds to the comparison function for determining whether to correspond, the selection function for selecting the category assigned to the demand pattern determined to correspond in the comparison step for each of the plurality of demand actual information, and the demand actual information. To realize the selected division and the achievement table setting function for setting the achievement table to which the division pattern which is the information of the electric power demand associated with the division is assigned for the predetermined period. It is characterized by.

According to the power demand forecasting device according to the first aspect of the present invention, the power demand forecasting method according to the second aspect, and the program according to the third aspect, the measurement was performed as a table used for forecasting the power demand. It is possible to create a performance table created based on the information of power demand. By predicting the power demand of the facility using this performance table, it is possible to predict the power demand without depending on the information in the calendar.

As the classification, a method of dividing such as weekdays and holidays can be exemplified, and further, weekdays may be further divided into first weekday, second weekday, and the like. Further, as the length of the predetermined period, 24 hours, 12 hours, 6 hours and the like can be exemplified. Further, the prediction table has a predetermined period corresponding to the number of days in one week, a predetermined period corresponding to the number of days in two weeks, and a predetermined period corresponding to the number of days in January. Things can be exemplified.

In the first aspect of the present invention, the division of the forecast period, which is the predetermined period for forecasting the power demand, is obtained based on the actual table, and the division is related to the division associated with the predetermined period. A forecast table that estimates the category in the forecast period based on the category information, and assigns the category based on the category information to the forecast period when the category based on the actual table and the category based on the category information are different. It is preferable that a prediction table setting unit for setting is further provided. Further, it is preferable to acquire the classification information associated with the predetermined period from the outside.

In the second aspect of the present invention, the division of the forecast period, which is the predetermined period for forecasting the power demand, is obtained based on the actual table, and the division is related to the division associated with the predetermined period. The forecast table that estimates the category in the forecast period based on the category information, and assigns the category based on the category information to the forecast period when the category based on the actual table and the category based on the category information are different. It is preferable to further have a prediction table setting step for setting. Further, it is preferable to acquire the classification information associated with the predetermined period from the outside.

By doing so, as a table used for forecasting the power demand, it is possible to create a forecast table obtained by modifying the actual table created based on the measured power demand information. For example, when the prediction interval includes an event that is difficult to reflect from the actual table, a prediction table that reflects the event can be created based on the classification information corresponding to the event. Therefore, it becomes easier to secure the prediction accuracy by predicting the power demand based on the prediction table as compared with the case based on the actual table.

According to the electric power demand forecasting device, the electric power demand forecasting method and the program of the present invention, the electric power demand forecasting accuracy is obtained by creating an actual table based on the measured electric power demand information as a table used for the electric power demand forecasting. It has the effect of making it easier to secure.

It is a block diagram explaining the structure of the electric power demand forecasting apparatus which concerns on one Embodiment of this invention. It is a flowchart explaining the arithmetic process which creates the performance table and the forecast table in the electric power demand forecasting apparatus of FIG. It is a schematic diagram explaining a plurality of classified demand patterns. It is a schematic diagram which shows the table corresponding to the January part to which the demand pattern is assigned, and the actual table corresponding to the January part to which a division pattern is assigned. It is a schematic diagram which shows the prediction table set by the prediction table setting part. It is a flowchart explaining the update process of calendar information.

The electric power demand forecasting device 10 according to the embodiment of the present invention will be described with reference to FIGS. 1 to 6. The electric power demand forecasting device 10 predicts the electric power demand in the retail electricity business. In the present embodiment, the power demand forecasting device 10 will be described by applying it to an example in which the power demand forecasting device 10 is an information processing device having a CPU (central processing unit), a ROM, a RAM, an input / output interface, and the like.

The program stored in the storage device such as the ROM described above is the acquisition unit 11, the storage unit 12, the analysis unit 13, the division unit 14, the comparison unit 15, and the selection unit 15 in cooperation with the CPU, ROM, RAM, and input / output interface. It functions as a unit 16, a performance table setting unit 17, and a prediction table setting unit 18.

The acquisition unit 11 is connected to the input unit 51, the external system 60, and the equipment 70 so that information can be communicated with each other. Here, the input unit 51 inputs the information of the user set value, which is the information appropriately set by the operator of the retail electric power business, to the electric power demand forecasting device 10. The acquisition unit 11 is connected so that the information of the user set value can be input from the input unit 51.

The external system 60 includes a customer management system 61, an information processing device 62 such as a server of an electric power company, and an information processing device 63 such as a server of a meteorological company. The customer management system 61 stores at least calendar information (classification information) including event information and holiday information for customers who receive electric power from the retail electric power business. In the information processing device 62 of the electric power company, at least, the demand record information, which is the information about the electric power supplied by the customer, is stored in association with the information of the date and time when the supply was received. The information processing device 63 of the meteorological company stores at least meteorological information which is weather and temperature information associated with date and time information. The input unit 51 is connected so that calendar information can be acquired from the customer management system 61, is connected so that demand record information can be acquired from the information processing device 62 of the electric power company, and can acquire weather information from the information processing device 63 of the meteorological company. It is connected to the.

The equipment 70 includes a database unit 71 that stores equipment information that is information about electrical equipment, and a database unit 72 that stores building information that is information about building equipment. The equipment information is information including the type and capacity of the electric equipment owned by the customer and using the electric power from the retail electric power business. Building information is information about a building in which electrical equipment is installed, and includes information such as age and floor area. The input unit 51 is connected so that equipment information can be acquired from the database unit 71 related to electrical equipment, and can be connected so that building information can be acquired from the database unit 72 related to building equipment.

The storage unit 12 stores various types of information acquired by the acquisition unit 11, and outputs information used in the processing when information processing is performed by the analysis unit 13 and the comparison unit 15. ..

The analysis unit 13 analyzes the power consumption pattern of the customer included in the performance information based on the performance information acquired from the information processing device 62 of the electric power company, and performs arithmetic processing to classify the power consumption patterns into a plurality of demand patterns. be.

Further, in the present embodiment, it is applied to an example of analyzing a consumption pattern, which is a pattern of power consumption of a customer for a predetermined period of one day (24 hours) and representing a change in consumption with the passage of time. I will explain. Further, in the arithmetic processing for analysis and classification, an example of performing arithmetic processing based on the consumption pattern included in the actual information based on seven actual information corresponding to one week will be described.

The predetermined period may be 24 hours as described above, may be longer than 24 hours, or may be shorter. Furthermore, the number of performance information used in the arithmetic processing for analysis and classification may be seven, which corresponds to one week, or may be the number corresponding to two weeks, which is longer than that, in January. It may be a corresponding number. Further, the number may be less than 7, which corresponds to one week.

The division unit 14 performs arithmetic processing for allocating a plurality of obtained demand patterns to predetermined weekday and holiday divisions. As the arithmetic processing for assigning, a known arithmetic processing method can be used and is not particularly limited.

The comparison unit 15 performs arithmetic processing for determining which of the plurality of demand patterns classified by the analysis unit 13 corresponds to the consumption pattern included in the demand record information.
The selection unit 16 performs arithmetic processing for selecting the division assigned to the demand pattern for the consumption pattern included in the demand record information associated with the demand pattern in the comparison unit 15.

The actual table setting unit 17 sets the actual table, which is a table to which the division selected by the selection unit 16 and the division pattern which is the information of the power demand associated with the division are assigned. The classification pattern is a predetermined pattern that represents a typical change in power consumption with the passage of time in the corresponding classification such as weekdays and holidays.

The forecast table setting unit 18 performs arithmetic processing for setting a forecast table to which a category based on the category information is assigned during the forecast period for forecasting the electric power demand. The details of the arithmetic processing for setting the prediction table will be described later.

Next, processing related to power demand forecasting in the power demand forecasting device 10 of the present embodiment will be described. First, the arithmetic processing for creating the actual table and the forecast table by the electric power demand forecasting device 10 will be described with reference to the flowchart of FIG.

As shown in FIG. 2, the acquisition unit 11 of the electric power demand forecasting device 10 performs a process of acquiring a plurality of demand actual information which is information on the past electric power demand measured in the last week (S101: Acquisition step). Specifically, it outputs a signal requesting the information processing device 62 of the electric power company to output a plurality of actual demand information measured in the last week, and is output from the information processing device 62 of the electric power company. The process of acquiring the plurality of demand record information is performed. The acquired plurality of demand record information is stored in the storage unit 12.

Next, the analysis unit 13 of the electric power demand forecasting device 10 performs a patterning process for obtaining a plurality of demand patterns based on the plurality of demand actual information acquired from the information processing apparatus 62 of the electric power company (S102: analysis step). Specifically, cluster analysis is performed based on the power consumption pattern of the customer included in the plurality of demand record information stored from the storage unit 12, and arithmetic processing for classifying into the plurality of demand patterns is performed. Here, as shown in FIG. 3, it will be described by applying it to an example of being classified into N types of demand patterns from pattern 1 to pattern N (N is a natural number).

As the calculation in which the analysis unit 13 analyzes the consumption pattern and classifies it into a plurality of demand patterns, a known analysis method such as cluster analysis can be used, and the analysis method is not particularly limited.

When a plurality of demand patterns are obtained, the division unit 14 of the electric power demand forecasting device 10 performs a naming process of assigning the obtained plurality of demand patterns to predetermined weekday or holiday divisions (S103: classification step). .. For example, a calculation process is performed in which a weekday division is assigned to the above-mentioned pattern 1 and the like, and a holiday division is assigned to the pattern 2 and the like.

In this embodiment, a plurality of demand patterns are assigned to two categories, weekdays and holidays, to be described. However, the weekday categories may be divided into three categories, the first weekday and the second weekday. You may increase the number of weekday divisions, or increase the number of holiday divisions.

Next, the comparison unit 15 of the power demand forecasting device 10 performs a comparison process of determining which of the plurality of demand patterns classified by the analysis unit 13 corresponds to the consumption pattern included in the acquired demand record information (S104: Comparison step). As the calculation process for determination, a known determination method such as a determination method based on the mean square error (RMSE) can be used.

When the demand pattern and the consumption pattern are associated with each other in the comparison unit 15, the selection unit 16 performs a pattern selection process for selecting the division assigned to the demand pattern for the consumption pattern (S105: selection step). ..

When the classification is selected, the performance table setting unit 17 links the classification selected in the selection unit 16 and the classification to the classification based on the date and time information linked to the consumption pattern included in the demand performance information. Performs the actual table setting process for setting the actual table, which is the table to which the division pattern which is the information of the power demand is assigned (S106: actual table setting step).

In the present embodiment, the example in which the actual table for January to which the division pattern shown as the lower table in FIG. 4 is assigned is set will be described. The upper table in FIG. 4 is a table corresponding to January to which the demand pattern is assigned. The performance table to be set may be assigned the division pattern for January as described above, or the division pattern for 1 to 3 weeks, which is less than that for January, is assigned. It may be the one to which the division pattern for several months is assigned.

When the actual table is set, the forecast table setting unit 18 performs a forecast table setting process for setting a forecast table to which a category based on the category information is assigned to the forecast period for forecasting the power demand (S107: forecast table setting step). ). Specifically, it is related to the arithmetic processing for obtaining the classification of the forecast period based on the performance table created by the performance table setting unit 17, and the classification associated with the prediction period, which is the calendar information including the prediction period. If the arithmetic processing that estimates the classification in the forecast period based on the classification information to be performed and the classification based on the actual table and the classification based on the classification information are different, set the prediction table that assigns the classification based on the classification information to the prediction period. Perform arithmetic processing.

Explaining with reference to FIG. 5, the prediction table setting unit 18 divides the next month (February), which is the prediction period, based on the January performance table created by the performance table setting unit 17 shown on the upper side of FIG. Performs arithmetic processing to obtain. Further, arithmetic processing is performed to estimate the appearance tendency (classification information) of weekdays and holidays in February from the appearance tendency (classification information) of weekdays and holidays based on the calendar information in January.

After that, the February classification calculated based on the January performance table is compared with the estimated February weekday and holiday appearance trends, and if they are different, the estimated February classification is compared. Performs arithmetic processing to set a prediction table to which the categories obtained from the appearance tendency of weekdays and holidays are assigned. In FIG. 5, the classification on the 3rd day of the end of February differs between weekdays and holidays, and a forecast table is set in which weekdays, which are the classifications obtained from the estimated tendency of appearance of weekdays and holidays in February, are assigned.

In addition, when obtaining the division of the next month which is the prediction interval used in S107, the calendar information obtained by the arithmetic processing described below may be used. First, as shown in the flowchart of FIG. 6, an arithmetic process for acquiring customer calendar information for the next month, which is a prediction interval, is performed from the customer management system 61 (S201).

An arithmetic process is performed to update the calendar information stored in the storage unit 12 to the customer calendar information acquired in S201 (S202).
When the update calculation process is completed, the prediction table setting process for setting the prediction table based on the calendar information stored in the storage unit 12 is performed (S107).

According to the power demand forecasting device 10 having the above configuration, it is possible to create a performance table created based on the measured power demand information as a table used for forecasting the power demand. By predicting the power demand of the facility using this performance table, it is possible to predict the power demand without depending on the information in the calendar.

Further, as a table used for forecasting the power demand, it is possible to create a forecast table obtained by modifying the actual table created based on the measured power demand information. For example, when the prediction interval includes an event that is difficult to reflect from the actual table, a prediction table that reflects the event can be created based on the classification information corresponding to the event. Therefore, it becomes easier to secure the prediction accuracy by predicting the power demand based on the prediction table as compared with the case based on the actual table.

10 ... Power demand forecasting device, 11 ... Acquisition unit, 13 ... Analysis unit, 14 ... Classification unit, 16 ... Selection unit, 17 ... Actual table setting unit, 18 ... Forecast table setting unit, S101 ... Acquisition step, S102 ... Analysis step , S103 ... Classification step, S104 ... Comparison step, S105 ... Selection step, S106 ... Actual table setting step, S107 ... Forecast table setting step

Claims (5)

An acquisition unit that acquires a plurality of actual demand information, which is information on electric power demand measured in a predetermined period, and
An analysis unit that obtains multiple demand patterns based on the acquired multiple demand record information,
A division unit that allocates the obtained plurality of demand patterns to a plurality of predetermined divisions, and
A comparison unit that determines which of the plurality of demand patterns each of the plurality of demand record information corresponds to.
For each of the plurality of demand record information, a selection unit for selecting the category assigned to the demand pattern determined to correspond to the comparison unit, and a selection unit.
A performance table setting unit that sets a performance table to which the selected classification and the classification pattern that is the information of the power demand associated with the classification are assigned to the predetermined period corresponding to the demand performance information. When,
Based on the performance table, a plurality of said categories of the forecast period, which is the predetermined period for forecasting the power demand, are obtained.
Based on the classification information associated with a plurality of said section is tied to the predetermined period, estimating a plurality of said section in the prediction period,
When the classification based on the performance table and the classification based on the classification information are different, the prediction table setting unit that sets the prediction table to which the classification based on the classification information is assigned to the prediction period, and the prediction table setting unit .
It characterized in that are provided power demand prediction unit. Wherein said sorting information that is tied to a predetermined time period, the power demand prediction apparatus according to claim 1, characterized in that the externally acquired. It is a method of forecasting power demand by a power demand forecasting device provided with an acquisition unit, an analysis unit, a division unit, a comparison unit, a selection unit, a performance table setting unit, and a forecast table setting unit.
An acquisition step in which the acquisition unit acquires a plurality of actual demand information, which is information on electric power demand measured in a predetermined period.
An analysis step in which the analysis unit obtains a plurality of demand patterns based on the plurality of acquired demand record information, and
A division step in which the division unit assigns the obtained plurality of demand patterns to a plurality of predetermined divisions.
A comparison step in which the comparison unit determines which of the plurality of demand patterns each of the plurality of demand record information corresponds to.
A selection step in which the selection unit selects the category assigned to the demand pattern determined to correspond in the comparison step for each of the plurality of demand record information.
The actual table setting unit assigns the selected division and the division pattern which is the information of the electric power demand associated with the division to the predetermined period corresponding to the demand actual information. Achievement table setting step to set
The forecast table setting unit obtains a plurality of the divisions of the forecast period, which is the predetermined period for forecasting the electric power demand, based on the actual table.
Based on the classification information associated with a plurality of said section is tied to the predetermined period, estimating a plurality of said section in the prediction period,
When the classification based on the performance table and the classification based on the classification information are different, the prediction table setting step for setting the prediction table to which the classification based on the classification information is assigned to the prediction period, and the prediction table setting step .
It is that power demand prediction method, comprising a. The power demand forecasting method according to claim 3 , wherein the classification information associated with the predetermined period is acquired from the outside. On the computer
An acquisition function that acquires multiple actual demand information, which is information on power demand measured during a predetermined period,
An analysis function that obtains multiple demand patterns based on the acquired multiple demand record information, and
A division function that assigns the required plurality of demand patterns to a plurality of predetermined divisions, and
A comparison function for determining which of the plurality of demand patterns each of the plurality of demand record information corresponds to, and
For each of the plurality of demand record information, a selection function for selecting the category assigned to the demand pattern determined to correspond in the comparison function, and a selection function.
A performance table setting function that sets a performance table to which the selected classification and the classification pattern which is the information of the power demand associated with the classification are assigned to the predetermined period corresponding to the demand performance information. When,
Based on the performance table, a plurality of said categories of the forecast period, which is the predetermined period for forecasting the power demand, are obtained.
Based on the classification information related to the plurality of divisions associated with the predetermined period, the plurality of the divisions in the prediction period are estimated.
When the classification based on the performance table and the classification based on the classification information are different, the prediction table setting function for setting the prediction table to which the classification based on the classification information is assigned to the prediction period, and the prediction table setting function.
A program characterized by realizing.

Images