神经网络是风电功率预测系统中应用最广泛的方法,而其训练算法是影响预测精度的重要因素之一。探讨了采用聚类法和正交最小二乘算法两种训练方法。以中国北方某风电场的实际数据以及数值天气预报数据为依据,对RBF聚类法和正交最小二乘算法进行了验证,最终研究并比较RBF不同预测情况与BP的差异。结果表明:对于提前24h的风电功率预测,RBF神经网络模型预测精度要好于BP神经网络模型,尤其以正交最小二乘算法为训练方法建立的RBF模型,预测精度较高,能够很好拟合实际功率曲线。 Neural network is the most widely used method in wind power prediction system, and its training algorithm is one of the important factors that affect the prediction accuracy. Two training methods using clustering method and orthogonal least squares algorithm are discussed. Based on the actual data of a wind farm in northern China and numerical weather forecast data, RBF clustering and orthogonal least squares algorithms are validated. Finally, the differences between RBF forecasting and BP are compared. The results show that the prediction accuracy of RBF neural network model is better than that of BP neural network model, especially for RBF model established by orthogonal least squares algorithm, which has high prediction accuracy and good fitting ability Actual power curve.