改进GM(1,1)预测模型对我国煤炭消费需求的预测分析  

Research on coal consumption prediction based on improved GM (1,1) prediction model

作　　者：周晓明 罗文柯[2,3] 李润球[2,3] 

机构地区：[1]湖南省白山坪矿业公司,湖南耒阳421828 [2]煤矿安全开采技术湖南省重点实验室,湖南湘潭411201 [3]湖南科技大学能源与安全工程学院,湖南湘潭411201

出　　处：《矿业工程研究》2010年第2期65-68,共4页Mineral Engineering Research

基　　金：国家安全生产监督管理总局安全生产科技发展计划项目(08-203;HN08-19;HNMJ10-05);湖南省教育科学"十一五"规划课题(XJY08BGD027);湖南省高校科技创新团队支持计划

摘　　要：为了预测未来煤炭消费需求状况,利用1998-2008年度我国煤炭消费需求的历史数据直接作为传统GM(1,1)及其残差模型的原始序列,通过生成处理后所得模型分别为勉强合格(三级)和合格(二级)等级,而通过对原始数据取自然对数为基础,并进行二阶弱化处理后所得的改进GM(1,1)模型及其残差GM(1,1)模型,经过点对点的残差检验发现,改进GM(1,1)模型及其残差GM(1,1)模型均提升至好的预测模型(一级)等级,其预测精度较高.用其预测未来3年的煤炭消费需求总量继续呈增长趋势,说明煤炭在未来短期内的主导地位没有改变.因此,国家和各级政府应加大对煤炭行业的资金投入与政策支持的力度,以保障我国经济持续稳定发展.To predict the coal consumption demand in China,the traditional GM（1,1） model,the traditional residual error GM（1,1） model,improved GM（1,1） model,and improved residual error GM（1,1） model were introduced and analyzed.Using the coal consumption from 1998 to 2008 as the original data sequence of traditional GM（1,1） and its residual error model,the results show that the tradition GM（1,1） model is the grade 3 qualified model and its residual error GM（1,1） model is the grade 2 qualified model.While the raw data by taking natural logarithm as data sequence of the improved GM（1,1） model and its residual error GM（1,1） model that improved by second-order weakening,based on residual error check of point-to-point,the result show that the two models improved the grade 1 qualified model and its prediction accuracy was more higher.Based on the improved predict model,the coal consumption demand in 2009,2010 and 2011is 1556.8,1744.2 and 1944.2 million tons respectively.The predict results show that the coal consumption continues to a growth trend and the dominated position of energy continues to a sustaining in the near future.So,the country and governments should increase financial input and strengthen policy support efforts in the coal industry to ensure a sustained and stable economy development in China.

关 键 词：煤炭资源 消费需求 改进 GM(1 1)模型 残差 预测 

分 类 号：X936[环境科学与工程—安全科学] TD173[矿业工程—矿山地质测量]