JP6642159B2 - Prediction support system, prediction support method and program - Google Patents

Description

  The present invention relates to a prediction support system, a prediction support method, and a program, and more particularly, to a prediction support system, a prediction support method, and a program that present future prediction values using past data.

  As described in Non-Patent Document 1, from 2016, the introduction of a licensing system in the electricity business and the introduction of a simultaneous plan value system are scheduled. The plan-value-equivalent system is such that each of the contractor (retail electricity supplier) and the power generation contractor matches the demand plan (or power generation plan) with the demand result (or power generation result) every 30 minutes. It is a system for coordination, which is one of the key factors in achieving a stable supply of electricity after liberalization.

  In general, it is known that the demand for electric power varies depending on factors such as weather, temperature, day of the week, and events performed in the area (see Background Art section of Patent Document 1). In addition, some methods have been proposed for obtaining a predicted value of power demand using various mathematical expressions and the like. For example, in Patent Literature 1, a power demand prediction support value that is a predicted value of power demand in the predetermined area is calculated based on an outside air enthalpy coefficient and a solar radiation coefficient calculated based on a temperature predicted value and a humidity predicted value. A power demand prediction support device including a power demand prediction support unit is disclosed.

JP 2015-97059 A

Electricity wide-area operation promotion organization, "Business briefing session on wide-area organization system", [online], [Search on March 7, 2016], Internet <URL: https://www.occto.or.jp/oshirase /hoka/files/koiki_sys_jigyoushasetsumeikai_shiryou_r3.pdf>

  The following analysis is provided by the present invention. The method is not limited to the prediction of power demand, but is a method of predicting the sales of specific products and visitors (stayers) to facilities as an approach different from automation such as the technology of Patent Document 1 based on past statistical data. Techniques for making future predictions are known. This method has a problem that although the prediction operation itself is relatively simple, it is difficult to select data to be used for prediction, and it is difficult to perform verification and feedback when a prediction error occurs. For example, when predicting tomorrow's power demand, prediction can be performed using past power demand data having the same characteristics of weather, temperature, day of the week, and the like. However, even if the weather, the temperature, the day of the week, etc. are the same, the prediction may be unsuccessful due to various factors, such as a closed day for some customers and an extended operation day for other customers. Similarly, when predicting the sales of a specific product, it is necessary to consider not only the weather, the temperature, and the day of the week, but also the presence or absence of an event in the region to be predicted, the advertisement execution schedule, and the like.

  An object of the present invention is to provide a prediction support system, a prediction support method, and a program that can contribute to improvement of prediction accuracy and easiness of verification in a method of performing future prediction from past statistical data.

  According to the first viewpoint, the first time series data obtained by actually measuring the time series change of the data belonging to the first category and the time series change of the data belonging to a second category different from the first category are described. Data storage means for storing the actually measured second time-series data, data selection means for receiving from the user a selection of a period to be used for prediction of each of the first and second time-series data, Predicted value presenting means for presenting, to the user, time-series data created using the first time-series data obtained and the selected second time-series data as a predicted value; When the selection operation of the time series data or the second time series data is received, the prediction related information of the first time series data or the second time series data used for creating the prediction value is transmitted to the user. And category data presenting means for presenting with al selected time period, the prediction assisting system comprising a are provided.

  According to the second viewpoint, the first time series data obtained by actually measuring the time series change of the data belonging to the first category and the time series change of the data belonging to a second category different from the first category are described. A data storage unit for storing the actually measured second time-series data, and receiving a selection of a common period common to the first and second time-series data from the user; First prediction value presenting means for creating the entire prediction data using the time-series data of the above, and presenting the prediction data to the user; Data selection means for receiving reselection, and presenting to the user, as predicted values, time-series data created using the selected first time-series data and the selected second time-series data. 2 Prediction support system and a predicted value presenting means.

  According to the third viewpoint, the first time series data obtained by actually measuring the time series change of the data belonging to the first category and the time series change of the data belonging to a second category different from the first category are described. A step of: receiving from a user a selection of a period to be used for prediction for each of the first and second time-series data, wherein the computer includes a data storage unit that stores the actually measured second time-series data. Presenting the time series data created using the selected first time series data and the selected second time series data to a user as a predicted value; When a selection operation of the time-series data or the second time-series data is performed, the prediction-related information of the first time-series data or the second time-series data used for creating the predicted value is previously stored. And presenting with the selected time period by the user, supporting estimation method comprising is provided. The method is tied to a specific machine such as a computer having the data storage means and a processor for performing various processes using the stored data.

  According to the fourth viewpoint, the first time series data obtained by actually measuring the time series change of data belonging to the first category and the time series change of data belonging to a second category different from the first category are described. Processing for receiving, from a user, a selection of a period used for prediction for each of the first and second time-series data, in a computer including a data storage unit that stores the actually measured second time-series data; A process of presenting, as a predicted value, time-series data created using the selected first time-series data and the selected second time-series data to a user; When receiving a selection operation of the time series data or the second time series data, the first time series data or the prediction related information of the second time series data used for creating the predicted value is transmitted to the user. Program for executing a process of presenting with al the selected time period, is provided. Note that this program can be recorded on a computer-readable (non-transient) storage medium. That is, the present invention can be embodied as a computer program product.

  ADVANTAGE OF THE INVENTION According to this invention, it becomes possible to improve the prediction accuracy and the verifiability in the method of making a future prediction from past statistical data.

It is a conceptual diagram for explaining composition and operation of one embodiment of the present invention. It is a conceptual diagram for explaining composition and operation of one embodiment of the present invention. It is a figure for explaining an effect of one embodiment of the present invention. 1 is a diagram illustrating a configuration of a power demand prediction support system according to a first embodiment of the present invention. FIG. 4 is a diagram illustrating an example of information held in a performance data storage unit of the power demand prediction support system according to the first embodiment of the present invention. FIG. 3 is a diagram illustrating an example of a hierarchical structure of performance data according to the first embodiment of the present invention. It is a figure showing another example of the hierarchical structure of the result data of a 1st embodiment of the present invention. FIG. 3 is a diagram illustrating an example of information stored in an event storage unit of the power demand prediction support system according to the first embodiment of the present invention. 3 is a flowchart illustrating an operation of the power demand prediction support system according to the first exemplary embodiment of the present invention. It is an example of the screen displayed on the terminal which accesses the electric power demand prediction support system of the first embodiment of the present invention. FIG. 11 is a diagram illustrating a state in which a date used for prediction is selected from the calendar in FIG. 10. FIG. 12 is a diagram illustrating a state in which prediction data is calculated using the selected day in FIG. 11. FIG. 13 is an enlarged view of a region 304 in FIG. 12. FIG. 13 is a diagram illustrating a state where a customer A selection button in FIG. 12 is pressed. It is a figure which shows an example of the ratio change screen of the consumer group A of FIG. FIG. 13 is a diagram illustrating a state where a customer B selection button in FIG. 12 is pressed. It is a figure which shows the example which changed the time series data used for the prediction of the consumer group A of FIG. FIG. 18 is a diagram illustrating an example of prediction data rewritten by the change in FIG. 17. It is an example of the screen displayed on the terminal which accesses the electric power demand prediction support system of the second embodiment of the present invention. FIG. 20 is a diagram illustrating a state in which a batch setting button is pressed after pressing a consumer selection button in FIG. 19. It is a figure showing an example of a batch change processing (bulk down) by a batch setting button. FIG. 22 is a diagram illustrating an example of prediction data rewritten by the batch change in FIG. 21. It is a conceptual diagram for explaining the modification composition and operation of one embodiment of the present invention.

  First, an outline of an embodiment of the present invention will be described with reference to the drawings. It should be noted that the reference numerals in the drawings attached to this outline are added to each element for convenience as an example to facilitate understanding, and are not intended to limit the present invention to the illustrated embodiments.

  According to one embodiment of the present invention, as shown in FIG. 1, a prediction support system including a data storage unit 11, a selection reception unit 12, a prediction value presentation unit 13, and a prediction target-specific data presentation unit 14. Can be realized.

  The data storage unit 11 measures first time series data of a data belonging to the first category and a time series change of data belonging to a second category different from the first category. And second time-series data. The first and second time-series data are obtained by dividing data having a similar tendency of change into different categories in the entire data to be predicted and totalizing the data, and recording the time-series change of the total value. Be composed. For example, when it is desired to predict a change over time in sales at a gas station, gasoline sales are totaled and stored as a first prediction target, and light oil sales are totaled and stored as a second prediction target. These past data are then used to predict future day-to-day changes in gas station sales, which are primarily made up of gasoline and gas oil sales.

  The selection receiving unit 12 receives, from the user, at least one or more selections of periods to be used for prediction with respect to the first and second time-series data. For example, in the example of FIG. 1, as the first time series data, data of March 5, 2016 and March 6, 2016 are selected. Similarly, as the second time series data, the data on March 6, 2016 and the data on March 13, 2016 are selected.

  The predicted value presenting means 13 is configured to include a time series configured to include a sum of an average of the selected two or more first time series data and an average of the selected two or more second time series data. The data is presented to the user as a predicted value. For example, in the example of FIG. 1, the average of March 5, 2016 and March 6, 2016 is calculated for the first time series data, and the average of March 6, 2016 is calculated for the second time series data. And the average of March 13, 2016 has been calculated. Then, by adding both, prediction data reflecting the tendency of each change of the first and second time-series data is obtained (see “Overall prediction value Pr1” in the lower right of the balloon in FIG. 1). . In the example of FIG. 1, the selection of two time series data is received for the first and second time series data to calculate the predicted value. However, the number of the time series data to receive the selection is reduced to two. There is no need to fix. For example, for data belonging to the first category, one period may be selected as long as good accuracy is obtained without selecting two periods. Alternatively, three or more periods may be selected, their average may be taken, and the predicted value may be calculated.

  In addition, in addition to the sum of the averages of the first and second time-series data, if there is an element to be considered (for example, sales of kerosene sales in cold regions), the value is added, A correction such as multiplying the sum of the second time-series data by a predetermined coefficient may be performed. For example, in addition to the sales of gasoline and diesel oil described above, if extraordinary sales that are not available in the past data (for example, sales of car supplies by a campaign) are expected on the forecast target date, add the number. It is desirable to perform processing.

  The prediction target-specific data presenting means 14 provides a function of confirming the “probability” of the predicted value calculated by the predicted value presenting means 13 and prompting reselection of data used for prediction as necessary. Specifically, the prediction target-specific data presentation means 14 receives the first time-series data or the second time-series data from the user and selects the first time-series data or the second time-series data. And the prediction-related information of the second time-series data or the second time-series data is presented together with a period selected by the user. For example, as shown in a balloon in FIG. 2, when the first time-series data constituting the overall prediction value Pr1 is selected, the prediction target-specific data presenting unit 14 uses March 2016 used for the calculation. The details of the first time-series data on the 5th and the March 6, 2016 are presented to the user. Similarly, when the second time-series data constituting the overall predicted value Pr1 is selected, the prediction target-specific data presentation unit 14 uses the data of March 6, 2016 and March 13, 2016 used for the calculation. The details of the second time-series data of the day are presented to the user. The mode of the prediction related information is not limited to the example of FIG. For example, if there is one selected time-series data, only the time-series data may be displayed, or other live-state information data on the day when the time-series data is created may be displayed together.

  According to the above-described prediction support system, it is possible to confirm the calculation basis of the entire predicted value presented by the predicted value presenting means 13 based on the information presented by the prediction target-specific data presenting means 14.

  Further, according to the prediction support system as described above, it is possible to confirm the “certainty” of the predicted value calculated as described above. For example, as shown in the upper graph of FIG. 3, a case is considered in which the predicted “overall predicted value Pr1” differs from the actual value of the same day of the previous year. The user can see from the amount of deviation that the “overall predicted value Pr1” is not appropriate. For example, as shown in the lower part of FIG. 3, the user reselects the data period used for the prediction of the second time-series data, recalculates the overall prediction value again, and determines the likelihood. You can check again.

  Further, according to the prediction support system as described above, the analysis of the cause of the occurrence of the misprediction and the like are facilitated by the same principle. When a prediction error as shown in the upper part of FIG. 3 occurs, the user notices that the time period used for the prediction is not appropriate when performing prediction next time under the same condition, and the first and second predictions are performed. After examining the similarity between the time series data and the period in which the prediction is performed, the periods used for the prediction are selected for the first and second time series data. That is, the above-described prediction support system is configured to improve the prediction accuracy by giving the user notice.

  In the above embodiment, as shown in FIG. 23, a configuration including a data storage unit 11, a first predicted value presenting unit 15, a selection receiving unit 12, and a (second) predicted value presenting unit 13 is provided. Can be transformed. The first predicted value presenting means 15 receives a selection of a common period common to the first and second time-series data from the user, and performs overall prediction using the first and second time-series data of the common period. Create data and present it to the user.

  According to the present embodiment, before the first predicted value presenting means 15 receives the data selection by the selection receiving means 12, as shown in the upper part of FIG. The whole prediction data (whole prediction value) is created using the series data, and presented to the user. Therefore, the selection of the period by the user in the selection receiving unit 12 is facilitated, and it is possible to easily obtain a highly accurate predicted value. Although the data presenting means for each prediction target 14 is omitted in FIG. 23, the data presenting means for each prediction target 14 can be provided also in this embodiment.

[First Embodiment]
Next, a first embodiment in which the present invention is applied to power demand prediction will be described in detail with reference to the drawings. FIG. 4 is a diagram showing a configuration of the power demand prediction support system 100 according to the first embodiment of the present invention. Referring to FIG. 4, actual data storage unit 110, event storage unit 111, date and time selection unit 120, actual data selection support unit 130, predicted data creation support unit 140, input unit 150, and output unit 160 , Is shown.

  The performance data storage means 110 corresponds to the data storage means 11 described above, and stores performance data (time-series data) of power demand per day and time in a service providing area of a contractor (retail electric power company). It is also preferable that the actual power demand data (time-series data) be stored in association with actual information at the date and time when these data were measured.

  FIG. 5 is a diagram illustrating an example of information stored in the performance data storage unit 110. Referring to FIG. 5, there is shown a table storing the actual values of weather, temperature, humidity, wind direction / wind speed, and power demand at each time, in addition to the day of the week on which the result data was created. Note that the weather, temperature, humidity, wind direction / wind speed are merely examples, and some of these may be omitted or other items that may affect power demand may be added. Good. Further, it is preferable that the power demand prediction support system 100 or the user obtains information to be stored in the performance data storage unit 110 from a server or the like on the network and stores the information in the performance data storage unit 110. In this way, the latest data is sequentially accumulated in the actual data storage unit 110.

  In addition, the actual value of the power consumption of the customer group in the area is stored below the power demand (total) column in FIG. For example, in these consumer groups, consumers who occupy a predetermined ratio with respect to the entire demand in the area are extracted, and consumers having similar consumption tendencies are grouped. In the example of FIG. 5, two customer groups A and B are created, and actual demand values for each day and time are stored. Examples of such consumer groups include companies belonging to a specific corporate group that have the same holidays and operating hours, and those that manufacture products with processes that use a lot of electricity during the manufacturing process. A group of companies and the like are included.

  FIG. 6 is a diagram showing a hierarchical structure of performance data including the customer groups A and B. In the example of FIG. 6, a customer group A and a customer group B are set in the area A. Of course, there may be a power consumer who does not belong to any customer group. Further, as shown in FIG. 7, the customer group may be further subdivided and some subgroups may be set. This is because even in the same corporate group, consumption trends such as holidays and operating hours may vary depending on the type of business. However, as can be seen by comparing FIGS. 6 and 7, an increase in the number of groups causes an increase in the number of categories to be predicted, thereby increasing the burden on the user. From this viewpoint, it is more effective to perform grouping of consumers such that a better prediction result can be obtained with a minimum number of groups, rather than increasing the number of groups unnecessarily. Needless to say, the minimum necessary group and sub-group may be required depending on the prediction target.

  The event storage unit 111 stores event information and the like useful for selecting past data for predicting power demand in a service providing area of a contractor (retail electric power company). FIG. 8 is a diagram illustrating an example of information stored in the event storage unit 111 of the power demand prediction support system. For example, according to FIG. 8, 2016/3/16 is recorded as a holiday of the customer A belonging to the customer group A. When the prediction target day falls on a holiday of the customer group A, by selecting the time series data of 2016/3/16 as one of the data used for the prediction, a good prediction result is obtained at least for the customer group A. Will be obtained. Similarly, 2016/3/18 in FIG. 8 is an all-day operation day of the customer B belonging to the customer group B. When the prediction target day corresponds to the all-day operation day of the customer group B, by selecting the time series data of 2016/3/18 as one of the data used for the prediction, at least the customer group B has a favorable prediction result. Is obtained. The event information may be input by the user, or the power demand prediction support system 100 may obtain the event information from a server or the like on a network and store the event information in the event storage unit 111.

  The date and time selecting unit 120 receives, via the input unit 150, the user's selection of the date and time at which the power demand is predicted. In a more preferred embodiment, the date and time selecting means 120 is preferably configured to prompt the user to select the earliest day (the day closest to the present time) among the dates of the demand plan to be created by the user. For example, when it is required to submit a demand plan for the next day, the date and time selection unit 120 sets the date of the next day to the initial value of the date selection column. By doing so, it is possible to suppress a user's date selection mistake.

  The performance data selection support means 130 includes a calendar display means 131 functioning as the above-described selection reception means 12, and a candidate date comparison graph display means 132. The calendar display unit 131 displays a calendar for selecting a date and accepts a user's selection of a candidate date (candidate date) of actual data used for prediction (for example, refer to an area 302 in FIG. 10). The candidate day comparison graph display unit 132 reads the data of the day (candidate day) selected by the calendar display unit 131 from the performance data storage unit 110, and displays a graph for comparing each candidate day ( (See area 303 in FIG. 10).

  At this time, the result data selection support unit 130 may read out the event of the day selected by the calendar display unit 131 from the event storage unit 111 and display the event (see the area 301 in FIG. 10).

  The prediction data creation support unit 140 includes a prediction data creation unit 141 and a category-specific data presentation unit 142. Specifically, the prediction data creation unit 141 reads, from the actual data storage unit 110, the time-series data on the date when the selection is received by the actual data selection support unit 130, and presents the data to the user. When there are two or more days on which the selection is accepted by the actual data selection support unit 130, the prediction data creation support unit 140 calculates the average of the time-series data and sets the average as the prediction data. Further, when there are two or more days when the selection is received by the actual data selection support unit 130, the prediction data creation unit 141 receives the input of the weight applied to each time series data, and calculates the weighted average. May be used as prediction data (for example, refer to “ratio” in area 305 in FIG. 10). The category-specific data presenting means 142 corresponds to the prediction target-specific data presenting means 14 described above. The details will be described later with reference to FIGS.

  Further, the forecast data creating means 141 creates a graph showing the time series change of the created forecast value, and displays the graph overlying the actual data of the day (candidate date) selected by the calendar display means 131. (See area 304 in FIG. 11). The user can confirm the validity of the predicted value by referring to the graph in addition to the calculated value.

  Upon receiving a prediction data determination instruction (output instruction) from the user, the prediction data creation unit 141 outputs the prediction data to a predetermined storage device via the output unit 160. A demand plan is created by allocating electric power generated by a contractor (retail electric power company) 's own power generation equipment or electric power procured from outside to the forecast data.

  The input means 150 and the output means 160 use a terminal or the like, or directly receive an instruction from a user who accesses the power demand prediction support system 100 and output a result of the front end application program for realizing an interactive operation. Composed of In the present embodiment, a description will be given assuming that information is presented to a user using the power demand prediction support screen illustrated in FIG. 10 and the like, and various instructions are received from the user.

  Each unit (processing means) of the power demand prediction support system 100 shown in FIG. 4 is implemented by a computer program that causes a computer constituting the power demand prediction support system 100 to execute the above-described processes using the hardware. It can also be achieved. Note that “means” used in the present embodiment can be replaced with “units”.

  Subsequently, the operation of the present embodiment will be described in detail with reference to the drawings. FIG. 9 is a flowchart illustrating an operation of the power demand prediction support system according to the first exemplary embodiment of the present invention. Referring to FIG. 9, the power demand prediction support system 100 receives a user's selection of a target date and time at which power demand is predicted (step S101).

  Next, the power demand prediction support system 100 reads and displays the event information of the selected date and time from the event storage unit 111. Further, the power demand prediction support system 100 displays a calendar including the selected date and time (step S102). Here, it is assumed that Wednesday, March 23, 2016 is selected.

  FIG. 10 is a diagram illustrating an example of the demand prediction support screen at the time of step S101. Referring to FIG. 10, in the area 301, event information on March 23, 2016 (Wednesday), which is the date and time selected in step S101, is displayed as target date information. In the area 302, a calendar for March 2016 to which March 23, 2016 (Wednesday) belongs is displayed. The user looks at such information, and the past date and time (candidate date) actual data (hereinafter, “candidate date”) that is likely to be useful for predicting the power demand on Wednesday, March 23, 2016 Is equivalent to the above-mentioned first and second time-series data periods.).

  Here, for example, when the user selects an arbitrary date (candidate day) of the calendar in the area 302, the power demand prediction support system 100 reads data of the corresponding day from the actual data storage unit 110 and displays the data in the area 303. (Steps S103 and S104). In the present embodiment, a plurality of dates (candidate dates) can be selected, and the data of the selected day is combined with the data of the target date and time to be predicted in the area 303 so that the data can be compared with the data of other days. Preferably, they are displayed in a superimposed or parallel manner.

  FIG. 11 and FIG. 12 show, in step S103, the past week and the two weeks immediately before that, which are likely to be useful for predicting the power demand on March 23, 2016 (Wednesday), March 16, 2016. It is a figure showing the state where day (Wed) and 9th (Wed) were selected. Here, it is preferable that the selected day is highlighted by the calendar display means 131 on the calendar in the area 302. In the example of FIG. 11, as an example of highlighting, the numbers of March 16 (Wed) and 9 (Wed), 2016 are surrounded by dotted lines.

  In the examples of FIGS. 11 and 12, the weather forecast data on March 23, 2016 (Wed), which is the prediction target date, and the weather forecast data on March 16, 2016 (Wed) and 9 (Wed), 2016 Is displayed. In the lower part of the area 303, the predicted temperature on Wednesday, March 23, 2016, which is the prediction target day, and the transition of the temperature on Wednesday, March 16, 2016 and Wednesday, March 9, 2016 are displayed. . In FIGS. 11 and 12, by switching the display of the drop-down list from “weather / temperature” to “temperature / humidity”, it is possible to display a graph for comparing the transition between the temperature and the humidity on each day. The user can select a day close to the condition of the prediction target day by referring to the area 303.

  Next, the power demand prediction support system 100 displays, in the area 305, actual data of the date and time (candidate date) selected by the user. Here, the power demand prediction support system 100 can receive an input of a ratio to be applied to the actual data of each day from the user (step S105). In the example of FIG. 11, since an input operation is not particularly performed, an initial value of “1” for data on March 16, 2016 and “1” for data on March 9, 2016 is input. .

  Thereafter, when the user clicks the calculation button 307 on the demand prediction support screen (step S106), the power demand prediction support system 100 creates the prediction data by applying the ratio received in step S105 to the actual data of each day. It is displayed (step S107).

  In parallel with the display of the forecast data, the power demand forecasting support system 100 displays the power demand by customer for the date (candidate date) selected in step S103 in the lower area 304 of the demand forecasting support screen in FIG. A graph for comparing and confirming the predicted data for each consumer obtained from the above and the predicted data indicating the power demand of the entire area obtained by summing up the predicted data for each consumer is displayed. The user corrects the prediction data in the area 305 while considering the graph displayed in the area 304, other events, weather, and the like.

  Here, a method of calculating the forecast data displayed in the lower area 304 of the demand forecast support screen of FIG. 12 will be described. The power demand prediction support system 100 reads the actual data of the power demand for each customer on the date (candidate date) selected in step S103 from the performance data storage unit 110, and creates prediction data for each customer. Here, since March 16 (Wed) and 9 (Wed), 2016 are selected, the power demand prediction support system 100 outputs the demand on March 16, (Wed) and 9 (Wed), 2016. Using the actual data of the house group A, the prediction data of the customer group A is created from the average value (see the predicted value of the customer A in FIG. 13). The power demand prediction support system 100 uses the actual data of the customer group B on March 16 (Wed) and 9 (Wed), 2016 to create the prediction data of the customer group B by the average value thereof ( The customer B predicted value in FIG. 13). Then, if necessary, the power demand of other consumers is added to the sum of the two to obtain prediction data of the entire area (see AVG of the entire area (consumer A + B) AVG in FIG. 13).

  When the comparison between the actual data and the prediction data in the area 304 is completed, the user clicks the prediction completion button 306 in FIG. Upon accepting the click operation of the prediction completion button 306, the power demand prediction support system 100 stores the predicted value in a predetermined storage device referred to by a device for creating a procurement plan (power allocation), which is a subsequent process. The power demand forecast data (demand plan data) is output (step S109). Note that the prediction data can be created by reselecting the date (review process) any number of times. For example, in the example of FIG. 9, by clicking the “Redo” button 308 in the upper part of FIG. Returning to the data correction operation, the prediction can be redone.

  In the case of performing the above-described redo, in this embodiment, by clicking the customer selection buttons 310a and 310b in FIG. 12, the category-specific data presenting means 142 uses the prediction data for each customer and the prediction data for each customer. It is possible to check the actual data by customer on the candidate day. FIG. 14 is an example of a graph displayed in the area 304 when the customer selection button 310a is clicked. In the example of FIG. 14, the actual data of the customer group A on March 16 (Wed) and 9 (Wed), 2016 are displayed together with the predicted data of the customer group A.

  The power demand forecasting support system 100 can change the ratio applied to the data of each candidate day when creating forecast data for each customer. FIG. 15 is a diagram showing an example of the ratio change screen 305a (corresponding to the weight receiving means) of the customer group A in FIG. In the example of FIG. 15, the ratio “2: 1” is entered in the customer-specific ratio column in the second row from the top. In this state, when the calculation button 307 is clicked, the prediction data for each customer can be corrected as shown in the rightmost column of FIG. Such a ratio change screen 305a may be displayed in the area 305 of FIG. 12, or may be separately displayed in a sub window.

  FIG. 16 is an example of a graph displayed in the area 304 when the customer selection button 310b is clicked. In the example of FIG. 16, the actual data of the customer group B on March 16 (Wed) and 9 (Wed), 2016 are displayed together with the predicted data of the customer group B. The user can confirm the “probability” of the predicted data for each customer by looking at such a graph showing the predicted data for each consumer and the actual data of the candidate day on which the calculation was based.

  If the user determines that the candidate date on which the calculation of the prediction data for each customer is not appropriate, the user can change the candidate date on which the calculation of the prediction data for each customer is based. Specifically, the power demand prediction support system 100 receives a user's reselection of candidate days for each customer via the calendar shown in the area 302 of FIG.

  For example, the user who feels that the prediction data of the entire area shown in FIG. 13 has a higher numerical value as a whole is not satisfied with the data of the customer group A shown in FIG. 14 on Wednesday, March 9, 2016. Can be changed to a day. FIG. 17 is a diagram illustrating a change in the graph when the candidate date used for the prediction of the customer group A is changed from Wednesday, March 9, 2016 to Tuesday, March 22, 2016. . As a result of changing the candidate date used for the prediction of the consumer group A from Wednesday, March 9, 2016 to Tuesday, March 22, 2016, the line indicating the predicted value of the consumer group A is downward. I'm moving.

  FIG. 18 is a diagram showing a change in the prediction data of the entire area A after the change of the customer group A prediction value. As shown in the figure, as a result of the downward correction of the line indicating the predicted value of the customer group A, the prediction data of the entire area A is corrected downward. In this way, the user can correct the prediction data by changing the candidate date. In addition, by saving the change of this candidate date and the ratio applied to the forecast data for each customer, it is possible to record what kind of candidate date the forecast data was created based on and how It is possible to leave.

  When the prediction data with “certainty” is obtained by the above review operation, the user clicks the prediction completion button 306 in FIG. Upon receiving the click operation of the prediction completion button 306, the power demand prediction support system 100 outputs the predicted data (demand plan data) of the predicted power demand to a predetermined storage device as described above (step S109). .

As described above, according to the present embodiment, basically, it is possible to accurately estimate the power demand only by selecting a plurality of candidate days under the same conditions as the day to be predicted. Furthermore, in the present embodiment, since it is possible to redo the prediction for each customer using FIGS. 13 to 18, the prediction is redone taking into account events and the like for each customer, and the result is ranked higher. Therefore, it is possible to reflect the result in the prediction data of the power demand for each area, which is the result of the above. Also, by separately recording the date and time used to create the prediction data and the ratio applied to them, the basis for creating the prediction data can be left. The basis for creating such forecast data should be used effectively for the contractor (retail electricity supplier) to obtain in-house agreement when submitting a demand plan and for the verification process when forecasts are not met. Can be.
.

  In the above-described embodiment, the actual data selection support unit 130 has been described as including the candidate date comparison graph display unit 132. A list of icons (for example, a time-series weather icon, an icon representing temperature in color) or the like that can compare day data may be displayed.

  In the above-described embodiment, the line graphs have been described as being superimposed on the areas 302 and 304. However, the display form of the graphs is not limited to this. For example, a bar graph may be used instead of a line graph, and a numerical value, another graph, or the like may be displayed together. For example, it is also possible to adopt a form in which a graph indicating a change in data on a past date and a graph indicating a change in data on a prediction target date are arranged vertically in two rows or left and right so that they can be compared. Further, for example, the average prediction data for the whole year or each month may be displayed together with the prediction data.

[Second embodiment]
Next, a second embodiment in which a batch correction function is added to the first embodiment will be described in detail with reference to the drawings. Since the second embodiment can be realized by a configuration substantially similar to that of the first embodiment, the following description will focus on the differences. FIG. 19 is an example of a screen displayed on a terminal accessing the power demand prediction support system according to the second embodiment of the present invention. The difference from the first embodiment shown in FIG. 12 and the like is that the customer selection buttons 310a / 310b are combined into one in the area 304, and the batch setting button 310c is provided. .

  FIG. 20 shows a graph displayed in the area 304 when the collective setting button 310c is clicked after clicking the customer selection buttons 310a / 310b, and a collective setting window 500. A field 501 in the collective setting window 500 is a numerical value input field. A drop-down box 502 in the collective setting window 500 is used for selecting correction contents such as addition, subtraction, multiplication, and division to prediction data using the value input in the numerical value input field. For example, as shown in FIG. 20, by putting the number "500" in the numerical value input field 501, selecting the kW subtraction, and clicking the execution button 503, the prediction data of the customer group A is shifted downward. Is possible.

  Further, similarly to the first embodiment, the downward correction of the prediction data of the customer group A as described above is also reflected in the prediction data of the entire area A as shown in FIG. Can be modified downward. Similarly, if the kW addition is selected, the prediction data of the customer group A can be shifted upward, and if the multiplication or division is selected, the prediction data of the customer group A is corrected to n times, 1 / n. Can be.

  Further, in the present embodiment, a time range to which the correction is applied may be selected from the prediction data. For example, when 3:00 to 15:00 in FIG. 21 is selected and the execution button 503 of the collective setting window 500 is clicked in a reversed state, the input correction content is applied only in the range of 3:00 to 15:00. You may do so.

  Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and further modifications, replacements, and adjustments may be made without departing from the basic technical idea of the present invention. Can be added. For example, the network configuration, the configuration of each element, and the expression form of a message illustrated in each drawing are examples for helping the understanding of the present invention, and are not limited to the configurations illustrated in these drawings.

  For example, in the above-described embodiment, an example in which the present invention is applied to the field of power demand prediction has been described. However, the present invention is applicable to any field that can perform prediction using past data. There are no restrictions. For example, the present invention can be preferably applied not only to power demand but also to fields in which prediction based on past data such as gas and water is effective. In addition, instead of the above-described actual data for each customer and each day, actual data aggregated in appropriate time units by category according to the data to be predicted is prepared. It can also be applied to prediction, prediction of the number of guests in facilities, and the like.

Finally, preferred embodiments of the present invention are summarized.
[First form]
(Refer to the prediction support system from the first viewpoint above)
[Second embodiment]
In the prediction support system according to the first embodiment,
Before accepting from the user a selection of a period used for prediction for each of the first and second time series data, accepting a selection of a common period common to the first and second time series data from the user, A prediction support system comprising: means for generating overall prediction data using first and second time-series data of a common period and presenting the data to the user.
[Third embodiment]
In the prediction support system according to the first or second aspect,
The data selection unit receives selection of one or more periods for at least one of the first time series data and the second time series data,
The prediction support system for presenting, as a predicted value, time-series data created using an average of the time-series data in which the one or more time periods are selected, to the user.
[Fourth embodiment]
In a third form of the prediction support system,
The prediction support system, wherein the category-based data presentation unit presents, as the prediction-related information, time-series data in which the one or more periods are selected, together with a period selected by the user.
[Fifth embodiment]
In the prediction support system according to the third or fourth aspect,
Furthermore, weight receiving means for receiving an input of a weight applied to at least one of the two or more time series data of the first time series data or the second time series data from the user,
The prediction support system that calculates the weighted average of the two or more time-series data using the weight value, and creates the prediction value.
[Sixth form]
In the prediction support system according to any one of the first to fifth aspects,
During the presentation of the prediction-related information of the first time-series data or the second time-series data by the category-based data presentation means,
The data selection unit receives a reselection of a period used for prediction for at least one of the first and second time-series data,
The prediction support system wherein the predicted value presenting means and the category-specific data presenting means recalculate a predicted value using the reselected time-series data and present the predicted value to the user.
[Seventh embodiment]
In the prediction support system according to any one of the third to sixth aspects,
During the presentation of the prediction-related information of the first time-series data or the second time-series data by the category-based data presentation means,
The data selection unit receives an instruction to correct at least one of the first and second time-series data,
The prediction support system, wherein the predicted value presenting means and the category-specific data presenting means recalculate a predicted value using the corrected time-series data and present the predicted value to the user.
[Eighth embodiment]
In a prediction support system according to a seventh aspect,
The prediction support system in which the correction is a process of uniformly adding or subtracting a predetermined value to or from the average of the time-series data.
[Ninth embodiment]
In the prediction support system according to any one of the first to eighth aspects,
The first and second time series data are daily time series data of energy consumption totaled for each customer in a certain area,
The prediction value presenting means, from a combination of past daily time series data of the customer in the certain area, to create time series data of the prediction target day,
The prediction support system for presenting the time-series data for each customer, which constitutes the time-series data of the prediction target date, by the category-based data presenting unit.
[Tenth embodiment]
(Refer to the prediction support method from the second viewpoint above)
[Eleventh form]
(Refer to the program from the third viewpoint above)
Note that the tenth to eleventh modes can be expanded to the second to ninth modes, similarly to the first mode.

  The disclosures of the above patent documents and non-patent documents are incorporated herein by reference. Modifications and adjustments of the embodiments or examples are possible within the framework of the entire disclosure (including the claims) of the present invention and based on the basic technical concept thereof. Various combinations or selections of various disclosed elements (including each element of each claim, each element of each embodiment or example, each element of each drawing, and the like) are possible within the scope of the disclosure of the present invention. It is. That is, the present invention of course includes various variations and modifications that could be made by those skilled in the art according to the entire disclosure including the claims and the technical idea. In particular, with respect to the numerical ranges described in this document, any numerical values or small ranges included in the ranges should be interpreted as being specifically described even if not otherwise specified.

DESCRIPTION OF SYMBOLS 11 Data storage means 12 Selection accepting means 13 Predicted value presenting means 14 Predicted object-specific data presenting means 15 First predicted value presenting means 100, 100a Power demand prediction support system 110 Actual data storing means 111 Event storing means 120 Date and time selecting means 130 Result data selection support means 131 Calendar display means 132 Candidate day comparison graph display means 133 Event display means 140 Prediction data creation support means 141 Prediction data creation means 142 Category-specific data presentation means 150 Input means 160 Output means

Claims (12)

First time-series data in which a time-series change of data belonging to the first category is actually measured, and second time-series data in which a time-series change of data belonging to a second category different from the first category is actually measured. And data storage means for storing
Data selection means for receiving from the user a selection of a period to be used for prediction for each of the first and second time series data;
Predicted value presenting means for presenting, to the user, time-series data created using the selected first time-series data and the selected second time-series data, as a predicted value;
When the user receives an operation of selecting the first time-series data or the second time-series data, the first time-series data or the second time-series data used to create the predicted value Category-based data presenting means for presenting prediction-related information together with a period selected by the user,
Prediction support system equipped with.   Before accepting from the user a selection of a period used for prediction for each of the first and second time series data, accepting a selection of a common period common to the first and second time series data from the user, The prediction support system according to claim 1, further comprising: a unit that generates entire prediction data using the first and second time-series data of a common period and presents the prediction data to the user. The data selection unit receives selection of one or more periods for at least one of the first time series data and the second time series data,
The prediction support system according to claim 1, wherein the predicted value presenting unit presents time-series data created using an average of the time-series data in which the one or more periods are selected to a user as a predicted value.   4. The prediction support system according to claim 3, wherein the category-based data presenting unit presents, as the prediction-related information, time-series data in which the one or more periods are selected, together with a period selected by the user. Further, from the user, weight receiving means for receiving an input of a weight applied to two or more time series data of at least one of the first time series data or the second time series data,
5. The prediction support system according to claim 3, wherein the predicted value presenting unit calculates the weighted average of the two or more time-series data using the weight value to create the predicted value. During the presentation of the prediction-related information of the first time-series data or the second time-series data by the category-based data presentation means,
The data selection unit receives a reselection of a period used for prediction for at least one of the first and second time-series data,
The prediction support system according to any one of claims 1 to 5, wherein the predicted value presenting means and the category-specific data presenting means recalculate a predicted value using the reselected time-series data and present the predicted value to the user. During the presentation of the prediction-related information of the first time-series data or the second time-series data by the category-based data presentation means,
The data selection unit receives an instruction to correct at least one of the first and second time-series data,
The prediction support system according to any one of claims 3 to 6, wherein the predicted value presenting means and the category-specific data presenting means recalculate a predicted value using the corrected time-series data and present the predicted value to the user.   The prediction support system according to claim 7, wherein the correction is a process of uniformly adding or subtracting a predetermined value to an average of the time-series data. The first and second time series data are daily time series data of energy consumption totaled for each customer in a certain area,
The predicted value presenting means, from a combination of past daily time series data of the customer in the certain area, to create time series data of the prediction target day,
The prediction support system according to any one of claims 1 to 8, wherein the category-specific data presenting means presents time-series data for each customer that constitutes the time-series data for the prediction target day. First time-series data in which a time-series change of data belonging to the first category is actually measured, and second time-series data in which a time-series change of data belonging to a second category different from the first category is actually measured. And data storage means for storing
Receiving a selection of a common period common to the first and second time-series data from a user, creating overall prediction data using the first and second time-series data of the common period, and presenting the data to the user First predicted value presenting means for performing
Data selection means for receiving, from a user, a reselection of a period used for prediction for each of the first and second time-series data;
Second predicted value presenting means for presenting to the user, as the predicted value, the time-series data created using the selected first time-series data and the selected second time-series data,
Prediction support system equipped with. First time-series data in which a time-series change of data belonging to the first category is actually measured, and second time-series data in which a time-series change of data belonging to a second category different from the first category is actually measured. And a computer provided with data storage means for storing
Receiving from a user a selection of a period to be used for prediction for each of the first and second time-series data;
Presenting time-series data created using the selected first time-series data and the selected second time-series data to the user as a predicted value;
When the user receives an operation of selecting the first time-series data or the second time-series data, the first time-series data or the second time-series data used to create the predicted value Presenting prediction-related information together with a period selected by the user;
A prediction support method including: First time-series data in which a time-series change of data belonging to the first category is actually measured, and second time-series data in which a time-series change of data belonging to a second category different from the first category is actually measured. And a computer provided with data storage means for storing
Processing for receiving a selection of a period used for prediction from the user for the first and second time-series data,
A process of presenting, to the user, time-series data created using the selected first time-series data and the selected second time-series data, as a predicted value;
When the user receives an operation of selecting the first time-series data or the second time-series data, the first time-series data or the second time-series data used to create the predicted value A process of presenting prediction-related information together with a period selected by the user;
A program that executes

Images