An in-depth forecasting method of electricity market sales considering customer churn rate

Abstract

The invention provides an in-depth forecasting method of electricity market sales considering customer churn rate. Firstly, a calculating method of user loss rate in electric power market is given based on survival analysis, and the number of electric power purchasing users at a certain time in the future is forecasted. Then, the confidence network of consumer power consumption is constructed andcombined with the prediction of consumer number, and the depth prediction model of power sales is designed. Finally, an adaptive inertia weight algorithm is proposed to solve the model. The method provided by the invention obviously improves the forecasting accuracy of the electricity sales volume.

Description

technical field [0001] The invention relates to the technical field of electric power market information, and relates to a method for in-depth prediction of electric power market sales in consideration of user loss rate. Background technique [0002] Electricity sales forecast is the basis for power sales companies to formulate power purchase plans, the main content of long-term load forecasting, and the basis for economic operation. Therefore, the accuracy of electricity sales forecast is of great significance to promote the development of electricity sales companies and the construction of the power market. Accurate electricity sales forecast plays an extremely important role in the performance assessment of electricity sales companies, control of profit balance, and power marketing. However, with the development of the electricity market, competition has been introduced on the electricity sales side, and a new pattern of "multiple buyers-multiple sellers" has gradually fo...

AI Technical Summary

Problems solved by technology

[0003] The prediction accuracy of the overall electricity sales of electricity sales companies is not very ideal. The main reason is that the existing electricity sales forecast methods have the following shortcomings: Although users can choose different electricity sales companies according to their needs, there are few literatures on electricity sales. The company considers the user churn rate when forecasting electricity sales. Most of the research focuses on external uncertain factors to improve the prediction accuracy of electricity sales companies, but does not consider the number of users.

Moreover, the existing forecasting models all use simple neural network or linear regression, it is difficult to learn any complex rules in electricity sales data, and there are shortcomings such as too long training time, large amount of calculation, and difficult to obtain probability distribution.

Images