考虑局部条件特征的风电功率短期预测  

SHORT-TERM PREDICTION OF WIND POWER CONSIDERING LOCAL CONDITION FEATURES

作　　者：张家安[1,2] 黄晨旭 李志军 Zhang Jiaan;Huang Chenxu;Li Zhijun(State Key Laboratory of Reliability and Intelligence of Electrical Equipment,Hebei University of Technology,Tianjin 300401,China;School of Electrical Engineering,Hebei University of Technology,Tianjin 300401,China)

机构地区：[1]省部共建电工装备可靠性与智能化国家重点实验室(河北工业大学),天津300401 [2]河北工业大学电气工程学院,天津300401

出　　处：《太阳能学报》2024年第12期220-227,共8页Acta Energiae Solaris Sinica

基　　金：河北省自然科学基金(E2020202142)。

摘　　要：提出一种考虑局部条件特征的风电功率短期预测方法。首先,基于斯皮尔曼相关系数对局部条件因素与风力机功率的相关性进行分析,确定风速、风向和对风角度等为影响风电场功率短期预测准确度的关键因素;然后,基于广义极值分布分别对关键因素的分布参数进行估计,并构建平均波动系数指标描述各风力机间的参数差异性,基于K-means++算法对风力机进行聚类;最后,采用主成分分析(PCA)方法提取机群内各风力机功率的关键特征,并基于双向循环神经网络(BiGRU)对机群功率进行预测,进而累加获取风电场的预测功率。以华北某风电场运行数据为算例,验证该方法的有效性。A short-term prediction method of wind power considering local condition features is proposed.Firstly,based on the Spearman correlation coefficient,the correlation between local condition factors and wind turbine power is analyzed.Wind speed,wind direction together with yaw angle are selected as key factors.Then,the distribution parameters of key factors are estimated separately with the generalized extreme value distribution,and an average fluctuation coefficient index is constructed to describe the parameter differences between each wind turbine.The wind turbines are clustered into several groups with the K-means++algorithm.Finally,the key features of each wind turbine cluster are extracted with principal component analysis(PCA).Based on Bidirectional gated recurrent units(BiGRU),the power of the cluster is accurately predicted and accumulated.Taking the operation data of a wind farm in North China as an example,the effectiveness of this method is verified.

关 键 词：风电功率 预测 聚类分析 神经网络 特征提取 

分 类 号：TM614[电气工程—电力系统及自动化]