一种基于径向基神经网络的短期风电功率直接预测方法

提出了一种基于RBF神经网络的未来24 h风电功率直接预测方法。为克服传统聚类算法局部寻优的缺陷,基于模糊C-均值聚类算法,提出了一种将遗传算法、模拟退火算法和模式识别技术相结合的模糊聚类算法。基于某风电场的实测数据,采用所提出的模糊聚类算法和几种常用方法分别确定径向基函数的中心,并采用最小二乘法解决权值学习问题。预测结果表明了基于RBF神经网络的风电功率预测方法能够有效提高预测精度,且证明了所提出的模糊聚类算法的优越性。

Short-term wind power direct forecasting based on RBF neural network

A method based on RBF neural network to directly forecast wind power for the future 24 h is proposed. To overcome the defects of traditional clustering algorithm, a fuzzy clustering algorithm combining with genetic algorithm, simulated annealing algorithm and pattern recognition is proposed based on fuzzy C-means algorithm. The fuzzy clustering algorithm and some common methods are used to select the center of radial basis function, and the learning of weight is solved by Orthogonal Least Square (OLS), based on measured data. The results indicate that the method to forecast wind power can improve the prediction accuracy, and prove the superiority of integrated clustering algorithm.