Using fractional discrete Verhulst model to forecast Fujian's electricity consumption in China

Abstract

Electricity consumption is main energy consumption in China, which is an indicator to estimate a region's economic performance. Since predicting electricity consumption has assumed increasingly importance for economic development, a new fractional discrete Verhulst model is proposed to accurately predict electricity consumption. Firstly, according to the idea of discretization, take the reciprocal of original data. The newly generated data sequence is used as initial data to build model and the calculation formula of new model is deduced according to the modeling mechanism of grey prediction model. Secondly, the idea of data mining is applicated, three-quarters of data are selected as training set, and optimal value of order is determined by particle swarm optimization algorithm. Finally, the remaining data is regarded as testing set to validate performance of the model. The new model is applied to predict electricity consumption of Fujian, and compared with four competing models. Mean relative error and residual error, considered as indicators, are measured respectively to assess accuracy and stability of the model. Besides, electricity consumption in next three years is predicted. The results show that new model has better accuracy and stability, which provides scientific information to balance the relationship between supply and demand of electricity.