基于两级级联聚类的神经网络风电功率预测

WIND POWER PREDICTION BY CASCADED CLUSTERING METHOD AND WAVELET NEURAL NETWORK

对历史日及预测日的风速特性进行比较,提取综合相似程度高的历史日作为训练样本,对不同类别分别建立预测模型。采用两级级联的混合聚类算法实现相似数据的最优选择,并构建基于改进粒子群优化的小波神经网络模型预测风电功率。通过对中国西部某风电场的算例仿真,表明该方法能够有效识别样本数据的筛选,提高风电功率预测精度。

Accurate wind power prediction can improve the economic operation and safety management of power system. The accuracy of wind power forecast could be improved if the training samples have the similar variation with the predicting day. A cascaded hybrid clustering method which contained both Euclidean distance and cosine angle distance is proposed to extract the most similar training samples. A wavelet neural network based on improved particle swarm optimization(IPSO)is adopted to optimize the wind power output.The predicting strategy is applied to wind power forecast in a wind farm of west china,the results show that the similar samples are identified and the predicting accuracy is improved effectively.