摘要
在农村经济发展规划中,精确合理地预测未来我国农村居民用电量对农村电网发展规划、电网改造建设具有重要的意义。研究选取1995-2014年影响我国农村居民用电量的9个指标因素,依次采取常规BP神经网络法、主成分-BP神经网络法、主成分回归法对我国农村居民用电量进行仿真,然后依次比较以上预测分析方法的预测误差(即进行预测精度比较),最终确定主成分-BP神经网络为本研究的最优方法,并运用该方法预测我国2015-2020年农村居民的电力需求量。
In the planning of rural economic development,predicting the consumption of the rural power in China＇s rural residents＇ has a great economic and political significance on the development of rural power planning and power grid renovation construction.This study selected nine indicators of factors that effect China＇s rural power consumption from 1995-2014,using the conventional BP neural network prediction method,principal component-BP neural network prediction method,principal component regression forecast method to forecast the rural residents＇ consumption in China,then comparing the forecast error of the methods mentioned above（that is the prediction accuracy comparison）,ultimately determining the principal components-BP neural network is the optimal method of this study,and using this method to predict the electricity demand of rural residents from 2015 to 2020.
出处
《电力学报》
2016年第2期162-166,170,共6页
Journal of Electric Power
关键词
用电量预测
BP神经网络
主成分分析
回归分析
electricity consumption forecast
the BP neural network
principal component analysis
regression analysis
