基于粗糙集和RBF神经网络的风电场短期风速预测模型  

Short-term wind speed prediction for wind farms based on rough sets and RBF neural network model

作　　者：王莉[1,2] 王德明[1] 张广明[1] 周献中[2] 

机构地区：[1]南京工业大学自动化与电气工程学院,江苏南京210009 [2]南京大学工程管理学院,江苏南京210093

出　　处：《南京工业大学学报（自然科学版）》2011年第6期67-71,共5页Journal of Nanjing Tech University(Natural Science Edition)

基　　金：江苏省科技厅工业科技支撑计划资助项目(BE2009166)

摘　　要：结合粗糙集提出了一种RBF神经网络短期风速预测模型。采用粗糙集对预测模型的输入特征空间进行约简,找出对未来预测的风速具有主要影响的因素,以此作为RBF神经网络预测模型的输入变量;在RBF神经网络训练的过程中,采用在线滚动优化策略,将最新的样本加入训练集,从而使预测模型能够跟踪风速的最新变化。将提出的方法用于某风电场的1 h短期风速预测,仿真实验结果表明该方法具有结构简单、预测精度高的优点。A radical basis function（RBF） neural network model combined with rough sets was used to predict short-term wind speed.Rough sets were used to reduce input feature space so that the significant factors for wind speed prediction could be found as the input variables of RBF neural network prediction model.Online rolling optimization was adopted in training RBF neural network.The latest sample was added into the training sets,thus the prediction model could catch recent changes of wind speed.The proposed method was used to predict wind speed in 1 h.Simulation results showed that the method had advantages of simplicity and high precision.

关 键 词：风力发电 短期风速预测 粗糙集 RBF神经网络 

分 类 号：TM614[电气工程—电力系统及自动化]