A Forecasting Method of Electricity Sales Considering the User Churn Rate in a Power Market Environment

Abstract

In order to improve the accuracy of forecasts of the electricity sales of power sales companies, a depth forecast model of electricity sales based on the characteristics of the power market is proposed. First, based on survival analysis, the calculation method of the user churn rate in the electricity market is given, and the number of users at a certain moment in the future is predicted. Then, users’ electricity consumption that calculated by the deep belief network and the predicted quantity of users are combined to design a forecast model of electricity sales. Finally, the model is solved utilizing the weighting algorithm of adaptive inertia. The analysis of the example shows that the proposed method achieves a significant improvement in the accuracy of power sales forecasting.