双加权最小二乘支持向量机的短期风速预测  

Short-term Wind Speed Forecast Based on Double Weighted Least Squares Support Vector Machine Algorithm

作　　者：潘学萍[1] 史宇伟 张弛[1] 

机构地区：[1]河海大学能源与电气学院,南京210098

出　　处：《电力系统及其自动化学报》2014年第1期13-17,66,共6页Proceedings of the CSU-EPSA

基　　金：国家自然科学基金项目(51207045)

摘　　要：提出了双加权最小二乘支持向量机的短期风速预测方法。考虑到离预测点越远的历史风速数据对预测值的影响越弱,对训练样本中输入向量数据进行第1次加权,以体现不同元素对预测影响的差异。同时为区分训练样本的差异性,降低异常样本的干扰,对训练样本进行第2次加权。对双加权后的训练样本,采用加权最小二乘支持向量机模型进行预测,降低了对异常点的敏感度,实现了对不同样本的区别对待。根据某实测风速数据进行了风速预测,结果表明,所提方法能提高风速预测精度。Double weighted least squares support vector machine algorithm for short-term wind speed forecast is proposed in this paper.The training data are weighted firstly to reflect the effects of different data on the predicting value.The training samples are weighted again to distinguish the difference of training samples,which can reduce the influence of abnormal sample.With the double weighted training sample,the weighted least squares support vector machine （LS-SVM） algorithm is used to predict the hourly wind speed.The proposed method improves the sparse characteristic of LS-SVM and reduces the sensitivity of abnormal points.The prediction results show that the proposed method can improve the prediction accuracy.

关 键 词：风速预测 双加权方法 加权最小二乘支持向量机 短期预测 

分 类 号：TK81[动力工程及工程热物理—流体机械及工程]