基于Chebyshev多项式的神经网络中长期负荷预测研究
    点此下载全文
引用本文:李 莎, 曾喆昭.基于Chebyshev多项式的神经网络中长期负荷预测研究[J].经济数学,2015,(1):99-102
摘要点击次数: 686
全文下载次数: 68
作者单位
李 莎, 曾喆昭 (长沙理工大学 电气与信息工程学院湖南 长沙 410004) 
中文摘要:高精度负荷预测在提高电力系统的安全性和经济性方面有着极其重要的意义,而现有的负荷预测方法因参数有限,难以完全反映其内在规律,因而导致预测结果不够准确.为此提出了一种基于Chebyshev多项式神经网络模型的预测方法.该方法使用递推最小二乘法训练神经网络权值系数,以获得高精度的参数估计,从而实现Chebyshev多项式神经网络模型对负荷量的最优拟合,再利用训练好的Chebyshev多项式神经网络模型实现中长期负荷预测.研究结果表明,该方法能较好模拟负荷变化规律,有效提高了负荷预测精度,在电力系统负荷预测中有较大的应用价值.
中文关键词:Chebyshev多项式  神经网络  递推最小二乘法  负荷预测
 
The Medium and Long-term Load Forecasting Based on Chebyshev Polynomial Neural-network
Abstract:Accurate load forecasting has very important significance for improving power system security and economy. And the parameters of the existing methods are too little, which make the prediction result inaccurate and can't reflect the inherent law completely. Therefore, this paper proposed a new combination model based on Chebyshev polynomial neural-network (CPNN) model, which uses the recursive least square (RLS) method to train the neural network weight coefficient, so as to realize load data fitting based on CPNN model. Finally the forecasting for medium and long-term was computed by using trained CPNN model. Theoretical analysis shows that the new combination model fits the law of load development well and it helps to improve the forecasting accuracy with high practical value.
keywords:Chebyshev polynomial  neural-network  recursive least squares (RLS)  load forecasting
查看全文   查看/发表评论   下载pdf阅读器