基于分区建模的复杂山地风电功率预测

Wind Power Prediction in Complex Mountainous Region Based on Partition Modeling

受地表粗糙度、复杂地形等因素的影响,对山地风电场整体风电功率进行预测存在较大困难。为解决此问题,提出利用K-means聚类算法对风电机组进行分区建模,根据风电机组的预测精度以及风电机组与子区域的相关系数,选择每个子区域内的参考风电机组。采用BP神经网络分位数回归模型对参考风电机组的风电功率进行预测。基于参考风电机组的预测结果,利用核密度估计(KDE)和稀疏贝叶斯学习(SBL)形成的组合预测模型对子区域的风电功率进行预测,将子区域的预测结果相加即可得到整体区域功率预测结果。实验结果表明采用分区建模的方法可有效提升复杂山地风电场功率预测的准确性。

In order to solve the problem that the mountain wind farm is affected by factors such as surface roughness and complex terrain,the overall wind power prediction is difficult.The K-means clustering algorithm is used to model the wind turbine partition.According to the prediction accuracy of the wind turbine and the correlation coefficient between the wind turbine and the sub-region,the reference wind turbine in each sub-region is selected.The BP neural network quantile regression model is used to predict the wind power of the reference wind turbine.Based on the prediction results of reference wind turbines,a combined prediction model formed by kernel density estimation(KDE)and sparse Bayesian learning(SBL)is used to predict the wind power of sub regions.The overall regional power prediction results can be obtained by adding the prediction results of sub regions.The experimental results indicate that the method of partition modeling can effectively improve the accuracy of power prediction for complex mountain wind farms.