基于小波分析与神经网络组合法的短期风电功率预测模型.pdfVIP

基于小波分析和神经网络组合法的短期风电功率预测模型 1,2 1 1 1 于安兴 ，陈国初 ，徐余法 ，俞金寿 1. 上海电机学院 电气学院, 上海200000 2. 华东理工大学信息科学与工程学院, 上海200000 摘 要 ：短期风电功率预测对于电力系统的调度和安全稳定运行具有重要的影响。为提高预测的准确性，本 文提出一种基于小波分析法与人工神经网络法组合建模的预测方法。首先应用小波分析将原功率序列分解 得到内部规律较强的子序列，为有效去除高频噪声而不致丢失有用信息，再次引用小波分析对与噪声相关 的高层细节信号去噪，然后分别对平稳性较好的逼近信号、低层细节信号和去噪后的高层细节信号建立非 线性自回归动态神经网络预测模型，最后将各层预测结果重构即得功率预测结果。误差分析表明，与单纯 的神经网络法及无去噪处理的小波分析和神经网络组合法相比，预测精度明显提高。 关键词 ：短期风电功率预测；小波分析；人工神经网络；消噪 中图分类号：TH 16 文章编号：1005-9830 （2012 ）00-0000-00 Short-term Wind Power Prediction Model Based on Wavelet Analysis and Neural Network Combination Method 1,2 1 1 1 Yu An-xing , Chen Guo-chu , Xu Yu-fa , Yu Jin-shou 1. School of electrical engineering, Shanghai Dian Ji University, Shanghai 20000, China 2. College of Information Science and Engineering, East China University of Science and Technology, Shanghai 20000, China Abstract: Short-term wind power prediction has an important impact on the power system scheduling and stable operation. To improve the accuracy, this paper presents a combination modeling prediction method based on wavelet analysis and artificial neural network. At first, the original signal is broken down into different frequency bands with the use of wavelet analysis, effectively strengthening the internal laws of each sequence, in order to remove high frequency noise effectively and avoid the loss of useful information, quote wavelet analysis again to denoise the high-level detail signals associated with noise. Then establish Nonlinear Auto-regressive with Exogenous Inputs (NARX) neutral network prediction model respectively. Finally, reconstruct the predicting outcomes of different levels and get the fin