基于组合模型的能源需求预测

能源是人类生存和发展的重要物质基础,也是当今国际政治、经济、军事、外交关注的焦点。能源需求预测是合理制定能源规划的基础。能源需求预测的模型很多,总的来说,可以分为单一模型预测和组合模型预测。本文在分析几种常用单一模型的优缺点和适用范围的基础上,建立BP神经网络与灰色GM的优化组合模型,对江苏省未来十五年煤炭和石油的需求量进行预测。结果表明:①随着经济的发展,未来江苏省对煤炭和石油的需求量逐渐增加,其中煤炭从2008年的19 601.39万t标准煤增加到2020年的25 615.26万t标准煤,年均增长率为1.81%;石油从2008年的2 628.64万t标准煤增加到2020年的3 532.60万t标准煤,年均增长率为1.36%;②基于BP网络与GM(1,1)的组合模型克服了单一模型的缺点,实现了优化组合模型"过去一段时间内组合预测误差最小"的原则,且预测结果误差较小,不仅适用于能源的中长期预测,还可以推广到其他领域。

Energy Demand Forecasting Based on Combined Model

Energy is the basis of human's survival and development, and it is also the focus in the international political, military and diplomatic fields. Energy demand forecast is the basis for establishing energy program. At present, there are many energy demand forecasting models. In general, they can be divided into two forecasting models: one is the single forecasting model and the other is the combined model. Based on the analysis of advantages and disadvantages of some common single models and the applicable scale, one optimized combination forecast model which is composed of BP neural network and gray model is set up. At last, we apply this method to predict the demand for coal and oil of Jiangsu Province from 2007 to 2021. The article drew following conclusions: On the one hand, with the development of economy of Jiangsu Province, the demand of coal and oil will gradually increase.Coal increases from 196.013 9 tons in 2008 to 256.152 6 tons 2020, and oil increases from 26.286 4 tons in 2008 to 35.326 0 tons in 2020, with an annual increase of about 1.81% and 1.36% respectively. On the other hand, the combined model can overcome the shortcomings of single model, and realize the principle that the combined errors are the smallest over a period. Meanwhile, the combined model can not only be used to the mid-long term forecasting of energy, but also to extend to other areas due to its accurate forecasted results.