Forecast of rural electric energy substitution potential based on nonlinear regression-neural network

Abstract

Accurate prediction of electric energy substitution potential in rural areas is of great significance for accelerating rural electrification construction and promoting low-carbon rural development. In this paper, a neural network-nonlinear regression-based prediction model for electric energy substitution potential is established to predict and analyze the longterm electric energy substitution potential in rural areas. The paper first selects four key factors that affect the electric energy substitution potential: rural GDP, rural population growth rate, the proportion of rural power consumption and policy support measures. Then, the BP neural network is used to predict the value of the four key factors. And taking these factors as independent variables, the fitting polynomial algorithm is used to perform nonlinear fitting on the amount of electric energy substitution, so as to obtain the fitted predicted value of electric energy substitution. In order to improve the fitting accuracy, the paper further uses wavelet neural network to intelligently correct the fitting residual sequence, and realizes the effective prediction of electric energy substitution potential in rural areas.