基于机群划分和列昂节夫逆阵的风电场理论功率计算方法  

A Wind Farm Theoretical Power Calculation Method Based on Wind Turbine Grouping and Lengtev Inverse Matrix

作　　者：邢西龙 李睿 赖晓路[1] 肖碧涛[1] 刘承志 邵会学 XING Xilong;LI Rui;LAI Xiaolu;XIAO Bitao;LIU Chengzhi;SHAO Huixue(Guodian Nanjing Automation Co.,Ltd.,Nanjing 210000,China;Huadian New Energy Group Co.,Ltd.,Fuzhou 350000,China;Huadian New Energy Group Co.,Ltd.Anhui Branch,Hefei 230000,China)

机构地区：[1]国电南京自动化股份有限公司,江苏南京210000 [2]华电新能源集团股份有限公司,福建福州350000 [3]华电新能源集团股份有限公司安徽分公司,安徽合肥230000

出　　处：《电工技术》2024年第24期59-64,共6页Electric Engineering

基　　金：国家重点研发计划“深远海风电场集群系统优化与控制关键技术”(编号2023YFB4203200)。

摘　　要：风电场理论功率计算对于恢复限电时段等非正常功率数据、提升风功率预测准确性、进行电力交易结算等具有重要意义。提出一种基于机群划分和列昂节夫逆阵的风电场理论功率计算方法。通过风机运行机理对机组数据进行筛选;通过多重共线性检验和特征提取对全场机群进行划分。对于强相关机群,采用列昂节夫逆阵充分整合线性强相关机群的解释变量,克服传统多重共线性方法简单删减或整合解释变量致使参数估计误差较大的缺点,将每个变量所反映的数据信息和变量间的相互作用考虑进来,以提升参数估计精度。对于线性弱相关机群,由于地表状况、风向、气压等存在差异,将样本数据根据风向、气压等影响因子进行区间划分,构建单机理论功率计算模型,计算限功率等时段的理论功率。最后,通过实际算例验证方法的有效性,对强相关机群和弱相关机群建立独立非参数核回归模型,有效消除多重共线性影响,提升计算精度。Theoretical power calculation of wind farm plays an important role in restoring abnormal power data such as power-cut periods,improving the accuracy of wind power prediction,and conducting power trade settlement.This paper proposes a method for calculating the theoretical power of a wind farm based on wind turbine grouping and lengtev inverse matrix.The unit data is screened through the operation mechanism of the wind turbine;the whole fleet is divided through multicollinearity inspection and feature extraction.For strongly correlated clusters,the Leonjiev inverse matrix is used to fully integrate the explanatory variables of the linear strongly correlated clusters,to overcome the shortcomings of traditional multicollinearity methods that simply deleting or integrating explanatory variables will cause large parameter estimation errors,taking the data information reflected by each variable and the interaction between the variables into consideration to improve the accuracy of parameter estimation.When it comes to linearly weakly correlated cluster,due to the differences in surface conditions,wind direction and air pressure,the sample data are divided into intervals according to the influence factors such as wind direction and air pressure,so as to construct a single machine theoretical power calculation model to calculate the theoretical power of limited power period.Eventually,the effectiveness of the method is verified through actual calculation examples,and independent non-parametric kernel regression models are established for the strongly correlated clusters and weakly correlated clusters,which effectively eliminates the effects of multicollinearity and improves the calculation accuracy.

关 键 词：风电场 理论功率 机群划分 列昂节夫逆阵 多重共线性 非参数核回归模型 

分 类 号：TB611[一般工业技术—制冷工程]