摘要
针对风电功率预测时风速信息冗余,导致神经网络难以把握内在规律而影响训练效率的问题,选取最佳风速数据进行主成分分析,采用改进小波BP神经网络对风速进行预测。通过KMO和Bartlett球度双检验选取最佳风速数据,从而充分利用主成分分析法对风速数据进行提取以优化神经网络的输入,提高网络收敛速度和预测精度。通过某风电场风速数据仿真分析,与其他预测方法进行对比,结果表明该模型预测精度高、泛化性能好,验证了该预测方法的正确性和有效性。
Against the redundancy of wind speed information in wind power prediction that results in difficulties for neural network to get the inherent law and therefore affect the training efficiency, the optimal wind speed data were selected for principal component analysis （PCA）, while improved wavelet-BP neural network （WNN） was adopted for wind speed prediction. Based on KMO and Bartlett＇s test of sphericity, the optimal wind speed data were selected and then extracted using PCA to optimize the inputs of neural network, thus improving the convergence rate and prediction accuracy of the network. By comparing the simulation results on wind speed data of a wind farm with other methods, the wavelet-BP neural network model based on PCA is proved to have high prediction accuracy and good generalization performance, verifying the correctness and effectiveness of this method.
作者
解坤
张俊芳
Xie Kun Zhang Junfang(School of Automation, Nanjing University of Science and Technology, Nanjing 210094, China)
出处
《发电设备》
2017年第2期86-91,共6页
Power Equipment
基金
国家自然科学基金资助项目(51507080)
