摘要
为了解决短期电力负荷不同预测方法的预测角度片面性、预测精度差等问题,提出了基于小波神经网络(WNN)的组合预测模型.首先用小波神经网络预测模型和历史平均模型分别进行预测,然后再通过小波神经网络对两单一模型的预测值进行组合.相比BP神经网络组合模型,该组合预测模型的预测精度大大提高.该模型同时引入模糊聚类分析的方法选取组合模型的训练样本,减少了训练样本的冗余性,提高了预测模型的精度.
In order to solve the defect and the poor prediction accuracy of different prediction methods in short-term power load forecasting, the combination forecasting method based on wavelet neural network is pro-posed. First, the wavelet neural network prediction model and the historical average model are used to predict respectively. Then, the predicted views of the two models are combined by wavelet neural network. Compared with BP neural networks combination model, the prediction accuracy of this combination forecasting model is improved greatly. Fuzzy clustering analysis method is utilized in this paper to select training samples of the combined model, the training sample redundancy is reduced and the prediction accuracy is improved.
出处
《郑州大学学报（工学版）》
CAS
北大核心
2013年第1期78-81,共4页
Journal of Zhengzhou University（Engineering Science）
基金
河南省科技攻关计划资助重点项目(112102210100)
关键词
短期电力负荷预测
组合预测
小波神经网络
模糊聚类分析
short-term electric power load forecasting
combination forecasting
wavelet neural network
fuzz-y cluster analysis
