基于小波—BP神经网络的短期风电功率预测方法  

A Short-term Wind Power Prediction Method Based on Wavelet Decomposition and BP Neural Network

作　　者：师洪涛[1] 杨静玲[1] 丁茂生[2] 王金梅[3] 

机构地区：[1]北方民族大学,宁夏回族自治区银川市750021 [2]宁夏电力公司,宁夏回族自治区银川市750021 [3]宁夏大学,宁夏回族自治区银川市750021

出　　处：《电力系统自动化》2011年第16期44-48,共5页Automation of Electric Power Systems

基　　金：国家自然科学基金资助项目(50847042)~~

摘　　要：建立风电功率预测系统并提高其预测精度是大规模开发风电的关键技术之一。基于数值天气预报,建立了反向传播(BP)神经网络风电功率预测模型,并采用某风电场实际数据分析了影响该模型预测精度的因素。针对原始风速及功率序列日特性不明显、BP神经网络不能完全映射其特性的缺陷,提出了一种基于小波—BP神经网络的预测模型。该模型利用小波将风速与功率序列在不同尺度上进行分解,并使用多个BP神经网络对各频率分量进行预测,最后重构得到完整的预测结果。研究表明该模型可有效提高预测精度。Establishing the wind power prediction system and improving the prediction accuracy is one of the key techniques for exploiting wind power.Based on numerical weather prediction,a wind power prediction model using the back propagation（BP） neural network is proposed.Factors that affect the prediction accuracy are analyzed using actual data of a certain wind farm.In the light of inconspicuous day characteristic of the original wind speed and the failure of the BP neural network to completely map its power sequence,a prediction model based on wavelet-BP neural network is proposed.With the wavelet-BP neural network model,the wind speed and power sequence are decomposed into different scales.Then the sub-sequences of different frequency components are predicted using multiple BP neural networks.Finally,the output data of BP neural networks are reconstructed to obtain the complete wind power predicting results.It is shown by the research results that the prediction accuracy of wavelet-BP neural network is effectively improved.

关 键 词：风电功率预测 BP神经网络 小波变换 频率分解 

分 类 号：TM614[电气工程—电力系统及自动化]