基于支持向量回归机的中国煤炭长期需求预测

China's Long-term Coal Demand Forecasting by Support Vector Regression

本文运用支持向量回归机(SVR)对我国煤炭1985—2008年需求量进行模拟与仿真,预测我国2009—2020年煤炭长期需求,并模拟解释变量不同增长率下煤炭的需求量,最后根据预测结果给出政策选择。研究结果表明,在基本情景下中国煤炭需求从2009年的20.648亿吨,到2017年将出现峰值,达到24.592亿吨,随后缓慢下降至2020年的23.186亿吨。说明在现有的能源消费情景下,中国高速经济增长依然要严重依赖煤炭。同时还发现,经济增长是煤炭需求增长的主要原因,影响最大,是能源政策的一个重要影响变量;工业结构(第二产业)影响次之,即便是微调,也会对煤炭需求有较大的抑制作用;城市化也是影响煤炭需求的一个重要因素,但影响相对较小。

This paper utilizes support vector regression （SVR） to make the simulation of China＇s coal demand form 1985 to 2008, and then forecasts the long-term coal demand from 2009 to 2020. It also makes policy simulations on the coal demand under different growth rates for various variables and offers policy suggestions. The result shows that China＇s coal demand will increase from 2. 064 8 billion tons in 2009 to 2. 4592 billion tons in 2017, and then slowly decline to 2. 3186 billion tons in 2020. It indicates that under the existing structure of energy consumption, maintaining the high-speed economic growth rate must heav- ily rely on coal. It is also found that the rapid economic growth is the main cause of coal demand increase in China. The adjustment of industrial structure, even if it is minor, will have big inhibitory impact on coal demand. Urbanization is also an important influencing factor for coal demand, but with less effect.