| Summary: | This paper improves the combined model I, which directly adds the predicted value of the ARIMA model and the predicted value of the BP neural network model. The linear fitting of the ARIMA model and the nonlinear fitting of the BP model are taken as the independent variable, and the per capita coal consumption sequence is taken as the dependent variable. By multiple linear regression, a new combined model II is constructed. Based on the analysis of the change rule of China’s per capita coal consumption over the years, the combined model II is used to fit the per capita coal consumption from 2014 to 2018. The result shows that the fitting errors are 0.62%, 0.17%, 0.04%, 0.04% and 0.07%, respectively. Compared with the combined model I, the combined model II improves the prediction accuracy. Finally, the combined model II is used to predict China’s per capita coal consumption from 2019 to 2023, which are 2878 kg, 2893 kg, 2906 kg, 2919 kg and 2926 kg. It is concluded that the per capita coal consumption increases slightly but the growth rate slows down, which provides reference for relevant government departments to formulate reasonable energy development policies.
|
|---|