基于L-M优化BP神经网络的风电功率预测

The Capacity Prediction for the Wind Power Based on L-M Optimized BP Algorithm

摘要: 在传统BP算法的基础上，将Levenbery-Marquardt优化法与神经网络模型相结合的L-M优化BP算法进行了深入应用和分析。此方法与传统算法相比提高了系统的学习速度，加快了网络的收敛。针对某风电场58台机组额定功率为850 kw的风电机组20天(每15分钟一个预测点)的历史数据使用L-M算法优化下的前馈神经网络模型——BP神经网络模型进行了该风电场的实时预测，结果表明该方法在一定程度上更好的逼近了真实的曲线。

Abstract: Based on the traditional BP algorithm, combining Levenhery-Marquardt optimized algorithm and a neural network forecasting method，this paper put forward a L-M optimized BP algorithm. The algorithm quickens the train, improves stability. For the real power data of 58 wind turbines of some wind farm in somewhere, a real-time prediction has been made based on L-M optimized BP algorithm, and the result shows that the algorithm produces better results than traditional method.