摘要
针对湖南省化石能源供需格局新变化,本文基于滑动窗口法建立了PSO-BP神经网络模型,预测煤炭、原油和天然气三种化石能源的供需形势。根据预测结果,改进的PSO-BPNN模型能够很好地拟合湖南省或其他地区能源供需变化情况,且预测效果优于BPNN、ARIMA和GM(1,1)三种模型。通过对预测结果进行分析,发现虽然三种化石能源的需求增长速度逐渐减弱,但缺口量仍然很大;除无生产能力的原油和天然气的省内保障率为100%外,煤炭的省内保障率约为80%;随着天然气的投入使用,预估短期内湖南省化石能源供需形势将步入深刻调整阶段。最后,根据三种化石能源供需发展趋势,本文对预测结果进行了探讨并提出相关建议,为推动湖南省能源改革提供数据、理论支持。
In view of the new changes in the energy supply and demand pattern,this paper established a PSO-BP neural network model based on the sliding window method to predict the supply and demand situation of the three fossil energy sources of coal,crude oil and natural gas.According to the forecast results,PSO-BPNN model based on sliding window method can well predict the changes of energy supply and demand,the improved PSO-BPNN model can well fit the energy supply and demand changes in Hunan province or other regions.And it s obviously better than GA-BPNN,ARIMA and GM(1,1)model.By analyzing the results,it s found that although the demand growth of the three fossil energy sources is gradually weakening,the gap is still large.Except for crude oil and natural gas without production capacity,the provincial guarantee rate of coal is about 80%.With the use of natural gas,it is estimated that the supply and demand situation of fossil energy in Hunan province will enter a stage of profound adjustment in the short term.Finally,for the development trend of the supply and demand of three fossil energy sources,this paper also discusses the prediction results and puts forward relevant suggestions to provide data and theoretical support for the promotion of energy reform in Hunan province.
作者
李湘旗
苏婷
胡东滨
文明
欧亦兰
LI Xiangqi;SU Ting;HU Dongbin;WEN Ming;OU Yilan(Economic Research Institute of State Grid Hunan Electric Power Company,Changsha 410004,China;Business School,Central South University,Changsha 410083,China;Changsha Commerce&Tourism College,Changsha 410116,China)
出处
《中国矿业》
2021年第11期30-36,共7页
China Mining Magazine
基金
国网湖南省电力有限公司科技项目资助(编号:5216A218000P)
湖南省科技创新平台人才计划资助(编号:2019TP1053)。
关键词
能源供需
缺口
滑动窗口法
粒子群算法
BP神经网络
energy supply and demand
gap
sliding window
particle swarm optimization
BP neural network
