| China is the world’s largest producer and consumer of coal, which is not only the backbone of China’s industrial expansion, but also the main contributor of carbon dioxide. The process of accurately forecasting the coal demand plays an important role in developing strategic policies and formulating tactical plans. This paper applies cointegration test to check the power supply and GDP as input variables, and proposes a novel hybrid forecasting procedure called PSO transfer function-noise model (PTFN) which employs the PSO to optimize the coefficients of the transfer function-noise model. To validate PTFN’s effectiveness, its forecasting accuracy is compared with ones of two other models:the ordinary transfer function-noise model and traditional Grey system model. These three models are used to forecast the coal consumptions from2009to2010based on the past30years’data. Results demonstrate that PTFN improves forecasting accuracy significantly over the other two models, and illustrate how GDP and power supply explain coal consumptions. Then PTFN is utilized to forecast coal demands from2010to2020. It is expected to rise approximately8.7%annually. Finally, this paper offers several policy suggestions for reducing the coal consumption and carbon dioxide emissions in China. |
