风电场短期风速预测的MRA-SVM模型  

MRA-SVM Model of Forecasting for Wind Speed in Short-term

作　　者：杨亚兰[1] 徐耀良[1] 钟绍山[1] 谢江媛 

机构地区：[1]上海电力学院电力与自动化工程学院,上海200090

出　　处：《电力系统及其自动化学报》2014年第5期44-49,共6页Proceedings of the CSU-EPSA

摘　　要：为了提高风电场短期风速的预测精度,提出了基于多分辨率分析和支持向量机(MRA-SVM)的预测模型。模型以历史风速序列为输入对数据进行多分辨率分析,用支持向量机对分解后的单支序列分别回归预测,叠加各序列的预测结果即为最终预测值。通过对某风场10 d的实测风速进行分析,预测了未来4 h的风速。用均方根误差和平均绝对百分比误差对模型进行评价,与单一的SVM方法相比,提高了预测精度。实验证明,模型具有较强的风速预测能力,能普遍适用于风速的短期预测。In order to improve the accuracy in predicting short-term wind speed,a model based on multiresolution analysis （MRA） and support vector machine （SVM）is proposed.The input sequence of the model is historical wind speed.The data decompose into several single sequences for multiresolution analysis,and then the forecasting single sequence with SVM is summed up as the ultimate forecasting speed.Through the real wind data of past 10 days at site,the wind speeds in future 4 hours are predicted.In order to evaluate the performance of the proposed model,root mean square error and mean absolute percentage error are adopted.Compared with the model of SVM,the precision is proved.The practical function and general application of the model is verified with experiments.

关 键 词：短期风速预测 多分辨率分析 支持向量机 均方根误差 平均绝对百分比误差 

分 类 号：TM614[电气工程—电力系统及自动化]