摘要
针对电网布局与建设等要求,需要根据地区提供的历史数据对该地区未来的用电量进行比较合理的预测。为此,构建基于递推最小二乘法算法的多项式预测模型,并提出基于代数多项式神经网络预测方法。该方法以多项式拟合模型构建神经网络拓扑结构,以模型参数作为神经网络权值,以往年每个季度的用电量数据作为参考值,使用递推最小二乘法对神经网络权值进行训练以获得多项式模型参数。仿真结果表明,该方法不仅具有良好的拟合效果,而且也具有良好的预测功能,在电力系统用电量预测中具有较大的应用价值。
Aiming at the power grid layout and construction requirements,future electricity con-sumption can be reasonable predicted according to the historical data.Thus,the polynomial pre-diction model was set up with the recursive least squares algorithm,and the algebraic polynomial prediction method was proposed in this paper.With polynomial fitting model,the method con-structs neural network topology structure,and the structure parameters can be used as the weights of neural network.Using the recursive least squares method for training the weights of neural network, the method obtained polynomial model parameters.The simulation results showed that,this method has good fitting results and good prediction function,which has great application value in power consumption forecast.
出处
《电力科学与技术学报》
CAS
北大核心
2015年第1期34-40,共7页
Journal of Electric Power Science And Technology
基金
湖南省自然科学基金(11JJ6064)
湖南省科技计划项目(2011GK3122)
长沙理工大学开放基金(13kfjj07)
关键词
电力系统
用电量预测
递推最小二乘法
神经网络
power systems
electricity consumption forecast
recursive least square method
neural network
