摘要
为了提高风电功率预测的准确度,提出了一种基于对手竞争惩罚学习算法(rival penalized competitive learning,RPCL)优化径向基函数(radial basis function,RBF)神经网络的风电功率预测模型.首先通过RPCL确定网络隐含层神经元数目以及中心点初始值,然后由K均值聚类法确定隐含层神经元的中心点和宽度,最后通过最小均值算法确定隐含层神经元与输出层神经元之间的权值.仿真结果表明:此优化模型相较于传统RBF网络具有更高的准确性.
For increasing the accuracy of wind power forecasting, a rival penalized competitive learningbased radial basis function （RBF） neural network model was presented. Firstly the number of neuralnetwork hiddenlayernodes and its initial center values were determined by rival penalized competitivelearning. And then the width of RBF and the center values of network were identified accurately throughKmeans clustering. At last,appropriate weights of network were estimated by least mean square. Theforecasting result shows that the presented model can lead to more accurate forecasting compared with thetraditional neural network.
出处
《北京工业大学学报》
CAS
CSCD
北大核心
2016年第5期674-678,共5页
Journal of Beijing University of Technology
基金
湖北省自然科学基金资助项目(2015CFB586)
关键词
风电功率预测
对手竞争惩罚学习算法
RBF神经网络
K均值聚类
wind power forecasting
rival penalized competitive learning algorithm
radial basis functionneural network
Kmeans clustering
