KR-20260066423-A - POWER SALES BIG DATA CORRELATION ANALYSIS SYSTEM BASED ON DETECTION OF FLUCTUATION FACTORS IN POWER SALES FORECAST AND ITS OPERATION METHOD

KR20260066423AKR 20260066423 AKR20260066423 AKR 20260066423AKR-20260066423-A

Abstract

According to various embodiments, a power sales big data correlation analysis system based on detecting fluctuation factors in power sales forecasting comprises a statistical database, an automatic data acquisition and storage device, a power sales data consistency verification and analysis device, a power sales trend analysis device, a key variable correlation analysis device, a scenario generation and storage device, a prediction algorithm training data acquisition and storage device, and a processor. The processor acquires power sales forecasting related data through the statistical database, stores and preprocesses the power sales forecasting related data through the automatic data acquisition and storage device, verifies omissions and errors in the preprocessed power sales forecasting related data through the power sales data consistency verification and analysis device, outputs the prediction accuracy of the power sales volume forecasting algorithm through the power sales trend analysis device, derives correlation coefficients necessary for training the power sales volume forecasting algorithm by utilizing artificial intelligence prediction techniques through the key variable correlation analysis device, stores the results output through the power sales data consistency verification and analysis device, the power sales trend analysis device, and the key variable correlation analysis device as a scenario through the scenario generation and storage device, and the stored scenario through the prediction algorithm training data acquisition and storage device It can be set to output. According to various embodiments, a method of operation of a power sales big data correlation analysis system based on detecting fluctuation factors in power sales forecasting may include: a step of acquiring power sales forecasting data through a statistical database of the system; a step of storing and preprocessing the power sales forecasting data through an automatic data acquisition and storage device of the system; a step of verifying omissions and errors in the preprocessed power sales forecasting data through a power sales data consistency verification and analysis device of the system; a step of outputting the prediction accuracy of a power sales volume forecasting algorithm through a power sales trend analysis device of the system; a step of deriving correlation coefficients required for training a power sales volume forecasting algorithm by utilizing an artificial intelligence forecasting technique through a key variable correlation analysis device of the system; a step of storing the results output through the power sales data consistency verification and analysis device, the power sales trend analysis device, and the key variable correlation analysis device as a scenario through a scenario generation and storage device of the system; and a step of outputting the stored scenario through a prediction algorithm training data acquisition and storage device of the system. Various other embodiments are also possible.

Inventors

김시연
고종민
신지강
권성구

Assignees

한국전력공사

Dates

Publication Date: 20260512
Application Date: 20241104

Claims (7)

In a power sales big data correlation analysis system based on detecting fluctuation factors in power sales forecasts, statistical database, Automatic data acquisition and storage device, Power sales data consistency verification and analysis device, Power sales trend analysis device, Major variable correlation analysis device, Scenario creation and storage device, Prediction algorithm training data acquisition and storage device and, Includes a processor, The above processor is, Through the above statistical database, data related to power sales forecasting is obtained, and Through the above data automatic acquisition and storage device, the above power sales forecast related data is stored and preprocessed, and Through the power sales data consistency verification and analysis device described above, omissions and errors in the preprocessed power sales forecast-related data are verified, and Through the above-mentioned power sales trend analysis device, the prediction accuracy of the power sales volume prediction algorithm is output, and Through the above-mentioned principal variable correlation analysis device, correlation coefficients required for training the power sales volume prediction algorithm are derived by utilizing artificial intelligence prediction techniques, and Through the above-mentioned scenario generation and storage device, the results output through the above-mentioned power sales data consistency verification and analysis device, the above-mentioned power sales trend analysis device, and the above-mentioned major variable correlation analysis device are stored as scenarios, and Through the above prediction algorithm training data acquisition and storage device, configured to output the stored scenario, System.
In Article 1, The above data automatic acquisition and storage device includes a scheduler, a data automatic acquisition device, a data storage device, and a data output device, and The above processor is, Through the above scheduler, signals for acquiring and outputting data related to the power sales forecast are generated according to the schedule required for analysis, and Through the above data automatic acquisition device, necessary information is acquired from the statistical database based on the data acquisition signal generated through the scheduler, and Through the above data storage device, the necessary information acquired is stored, and Set to output the necessary information from the data storage device to the power sales data consistency verification and analysis device and the scenario generation and storage device based on the data output signal generated through the scheduler through the data output device, System.
In Paragraph 2, The above-mentioned power sales data consistency verification and analysis device includes a power sales loss rate statistical analysis device, a power sales volume anomaly detection analysis device, a monthly/monthly key variable and performance correlation analysis device, a power sales data consistency analysis and notification device, a power sales volume peculiar data restoration device, and a power sales consistency scenario comparison analysis device. The above processor is, Through the above-mentioned power sales loss rate statistical analysis device, statistical analysis data obtained by performing statistical analysis based on the above-mentioned necessary information output through the above-mentioned automatic data acquisition and storage device is output to the above-mentioned power sales volume anomaly detection analysis device, and Through the above-mentioned power sales volume anomaly detection and analysis device, a specific date on which at least one anomaly occurred is output based on the above-mentioned statistical analysis data, and Through the above-mentioned monthly/monthly key variable and performance correlation analysis device, the pattern of key variables and performance correlation of the above-mentioned specific date are analyzed, and a composite index is calculated based on the above-mentioned pattern and performance correlation. Through the power sales data consistency analysis and notification device described above, based on the comprehensive index, a notification is output on a date corresponding to the specific date on which the abnormal phenomenon occurred, and Through the above-mentioned power sales volume peculiar data recovery device, power sales volume data for the specific date on which the above-mentioned abnormal phenomenon occurred and the above-mentioned notification was output is deleted and restored, and Through the power sales consistency scenario comparison analysis device described above, the device is configured to query performance data and loss rates for a pre-set specific period, save and analyze the performance data and loss rates for the pre-set specific period as a scenario, and compare and analyze the scenario with another previously generated scenario. System.
In Paragraph 3, The above-mentioned power sales trend analysis device includes a power sales volume trend analysis device, a power sales volume maximum/minimum analysis device, and a power sales volume similar year/month analysis device, and The above processor is, Through the above-mentioned power sales volume trend analysis device, power sales volume trend data corresponding to the search condition is analyzed based on the first search condition set by the user, and Through the above-mentioned maximum/minimum power sales analysis device, power sales performance data corresponding to the above-mentioned first inquiry condition is output, and Through the power sales volume similar year/month analysis device described above, configured to analyze similar data corresponding to the first inquiry condition, System.
In Paragraph 4, The above-mentioned major variable correlation analysis device includes a weather information correlation analysis device and a major indicator variable correlation analysis device, and The above processor is, Through the above weather information correlation analysis device, the weather information correlation corresponding to the second search condition set by the user is analyzed, and Through the above-mentioned major indicator variable correlation analysis device, configured to analyze the major indicator correlation corresponding to the third search condition set by the user, System.
In Paragraph 5, The above scenario generation and storage device includes a power sales data consistency scenario generation device, a power sales trend analysis scenario generation device, a major variable correlation analysis scenario generation device, a standard management device, and a scenario storage device. The above processor is, Through the power sales data consistency scenario generation device, monthly power sales volume scenario data corresponding to a specific date on which the anomaly occurred is generated and output to the scenario storage device. Through the above-mentioned power sales trend analysis scenario generation device, scenario data for a period corresponding to a preset specific period is generated and output to the above-mentioned scenario storage device, and Through the above-mentioned main variable correlation analysis scenario generation device, scenarios are generated by dividing them into multiple pre-set groups and output to the above-mentioned scenario storage device, and Through the above standard management device, the abnormality of the data output to the above scenario storage device is output, and Through the above-mentioned scenario storage device, the above-mentioned output data is stored and configured to be output to the above-mentioned prediction algorithm training data acquisition and storage device, System.
In the operation method of a power sales big data correlation analysis system based on detecting fluctuation factors in power sales forecasts, A step of acquiring data related to power sales forecasting through the statistical database of the above system; A step of storing and preprocessing the power sales forecast-related data through the automatic data acquisition and storage device of the above system; A step of verifying omissions and errors in the preprocessed power sales prediction-related data through a power sales data consistency verification and analysis device of the above system; A step of outputting the prediction accuracy of a power sales volume prediction algorithm through a power sales trend analysis device of the above system; A step of deriving correlation coefficients necessary for learning a power sales volume prediction algorithm by utilizing an artificial intelligence prediction technique through a principal variable correlation analysis device of the above system; A step of storing the results output through the power sales data consistency verification and analysis device, the power sales trend analysis device, and the major variable correlation analysis device as a scenario through the scenario generation and storage device of the above system; and A step of outputting the stored scenario through the prediction algorithm learning data acquisition and storage device of the above system; including, How the system operates.

Description

Power Sales Big Data Correlation Analysis System Based on Detection of Fluctuation Factors in Power Sales Forecast and Method of Operation The present invention relates to a power sales big data correlation analysis system based on detecting fluctuation factors in power sales forecasts and a method of operation thereof. More specifically, it relates to an automated correlation analysis system and a method of operation thereof that utilizes highly reliable big data analysis technologies, such as data consistency verification, trend analysis, and correlation analysis, in real time, and enables the analyzed data to be subdivided and trained by region, contract, and usage into a deep neural network model of the latest artificial intelligence techniques. Decision-making related to business activities begins with forecasting activities, and forecasting activities are fundamental activities that improve management efficiency by reducing uncertainty and risk burden in all management activities. Electricity sales forecasting is critical as it significantly impacts the stable operation and management of the power system. If errors in forecasting cause difficulties in establishing mid-to-long-term investment plans, it can lead to various undesirable consequences for companies, such as reduced stability of the entire power system due to power supply shortages and revenue losses resulting from overinvestment in facilities. (Annual savings of 1.95 billion KRW in operation and supply costs due to a 1% improvement in electricity sales forecasting accuracy) Furthermore, highly accurate short- and medium-term power sales forecast data is essential to promote distributed energy and strengthen regional and locational signals following the implementation of the Distributed Energy Act. Electricity sales volume fluctuates in the short term under the influence of climate conditions and is closely related to economic factors such as Gross Domestic Product (GDP), Gross National Income (GNI), and the Industrial Production Index, as well as demographic changes, in the medium to long term; however, the difficulty of forecasting is increasing recently due to the rise in variable factors affecting sales volume, such as new and renewable energy, electric vehicle charging, PPAs, and RE100. Medium-term electricity sales data serves as fundamental information for key mid-term decision-making, such as the formulation of corporate management strategies and investment plans; therefore, sophisticated sales (electricity sales volume) forecasting is necessary to improve management efficiency. Therefore, securing technology capable of accurately predicting changes in sales volume by reflecting not only macroeconomic indicators such as economic indicators and demographic structure but also factors affecting the power industry environment is a top priority task that must be addressed to respond to the rapidly changing power environment. <Lack of an automated system for forecasting short- and medium-term power sales> The current electricity sales forecasting model has been operating since 2004 using a regression model based on econometrics. Although the model has been improved every two years to reflect fluctuating factors, no further improvement activities have been carried out since 2015. The improved Electricity Sales Forecasting Model (ESFOS) in 2015 is a causal model that predicts a dependent variable (electricity sales volume) that fluctuates according to independent variables such as GDP, temperature, and effective days, and provided forecast information for up to 18 months for sales volume by contract type. However, due to recent abnormal climate conditions and uncertainty regarding future economic prospects, the accuracy of electricity sales forecasts is declining, and the short forecast period limits their use in establishing mid-to-long-term management strategies. Currently, the short-term sales volume forecasting system (ESFOS) provides forecast information up to 18 months as shown in Figure 1, but in order to utilize it for establishing mid-to-long-term business strategies, a forecasting system that provides forecast information up to 60 months is required. <Lack of Data Verification/Analysis Methods Required for Electricity Sales Forecasting> Currently, when establishing investment plans for stable power grid operation and power infrastructure construction, companies verify and analyze performance data on electricity sales and forecast future mid-to-short-term sales; however, there is a lack of methods to verify the factors influencing increases or decreases in sales or the analysis procedures involved. In particular, the performance of the latest artificial intelligence algorithms designed to improve the accuracy of electricity sales forecasts varies significantly depending on the extent to which they are trained by reflecting key variable factors highly correlated with accurate input data; therefore, technologies for verifying and anal