Energy is a complex system affected by multiple factors, accurate energy demand forecasts provide the basis for the formulation and implementation of energy planning. This paper builds a new model and predicts China’s energy consumption. This study drew three main conclusions. First, aco-integration test and Granger causality test can help users discover the relationships between China’s energy demand and its influencing factors. Second, the improved PSO-LSSVR model showed its superiority over other models in terms of forecasting energy demand, which further improved prediction accuracy. Third, the forecasting results indicate that China’s energy demand will peak in 2034, and that the peak is 6.7 billion tonnes of coal equivalent (tce). Based on the forecasting results, the paper offers suggestions related to China’s energy development policy.
Keywords Markov chain, Chinese energy demand forecasting, PSO-LSSVR algorithm, Co-integration test, Granger causality test