摘要
为准确预测电力消费并给电力发展规划制定提供依据,提出一种电力消费混合预测模型(PSOGRNN),将GDP、人均可支配收入和电力消费历史数据作为输入变量,运用粒子群优化(PSO)算法优化选择用于电力消费预测的广义回归神经网络(GRNN)模型参数值,以此提高模型的预测精度。实例验证结果表明,与自适应GRNN模型、DGM(1,1)模型和最小二乘线性回归模型相比,PSOGRNN模型的预测精度最高,且有效可行。
In order to forecast electricity consumption accurately and provide some references for the formulation of electric power development plan,a hybrid electricity consumption forecasting model（PSOGRNN） is proposed.Taken the GDP,disposable income per capita and historical data of electricity consumption as the input variables,the parameter value of GRNN model for electricity consumption forecasting is automatically selected by PSO so as to improve the forecasting accuracy.Compared with self-adaptive GRNN model,DGM（1,1） model and the least squares linear regression model,example validation result shows that PSOGRNN model has high forecasting accuracy,and this method is effective and feasible.
出处
《水电能源科学》
北大核心
2013年第2期221-223,共3页
Water Resources and Power
基金
国家自然科学基金资助项目(71071053)
关键词
电力消费
预测
广义回归神经网络模型
粒子群优化算法
参数优化
electricity consumption
forecasting
generalized regression neural network model
particle swarm optimization algorithm
parameter optimization
