高频激励下变压器绕组梯形等效网络参数辨识方法研究  

Ladder Network Parameter Identification of Transformer Winding Under High-frequency Excitation

作　　者：任富强 张鸿儒[1] 康朝阳 汲胜昌[2] 李清泉[1] REN Fuqiang;ZHANG Hongru;KANG Zhaoyang;JI Shengchang;LI Qingquan(Shandong Provincial Key Laboratory of UHV Transmission Technology and Equipment(Shandong University),Jinan 250061,Shandong Province,China;State Key Laboratory of Electrical Insulation and Power Equipment(Xi’an Jiaotong University),Xi’an 710049,Shaanxi Province,China)

机构地区：[1]特高压输变电技术与装备山东省重点实验室(山东大学),山东省济南市250061 [2]电力设备电气绝缘国家重点实验室(西安交通大学),陕西省西安市710049

出　　处：《中国电机工程学报》2023年第11期4463-4473,共11页Proceedings of the CSEE

基　　金：山东省自然科学基金项目(ZR2021QE094)。

摘　　要：梯形等效网络在变压器绕组变形诊断领域占据重要地位。该文探索基于绕组可测频响数据,构建参数辨识算法实现高频激励下梯形网络元件值的反演。首先,确定参数辨识所用驱动点导纳函数的测试方式,保证网络的建模精度及最简结构;其次,构建结合遗传算法与迭代算法的网络元件反演算法、基于元件电压电流关系的网络求解模型及基于绕组等效电感电容的元件搜索域计算方法;最后,提出实体变压器高频段频响曲线的确定方法,研究参数辨识的计算效率及优化精度。辨识所得实体变压器的等效网络元件值符合相关约束条件,在800kHz~2MHz频段内对应曲线与实测曲线具备极高的吻合度,其目标函数值从442.11下降至90.44,降幅高达79.54%,从而验证了该文所提方法的有效性和精确性。This paper focuses on the inversion of ladder network parameters for transformer windings in the context of diagnosing winding deformations.The goal is to utilize parameter identification algorithms and measurable frequency response analysis(FRA)data under high-frequency excitation.To begin with,the paper determines the test arrangement for the driving-point admittance(DPA)function,which is crucial for accurate modeling and achieving the simplest network topology.The combination of Genetic Algorithm and Iteration Algorithm is then employed to identify the network parameters.The network mathematical model based on the U-I relationship,and the feasible region of network parameters derived using the winding equivalent inductance and capacitance components,are systematically established.The paper also determines the high frequency(HF)region of the DPA data and evaluates the computation efficiency and accuracy of the proposed method.The obtained network parameters for an actual transformer meet all relevant constraints.Furthermore,the corresponding DPA curve within the frequency range of 800kHz~2MHz exhibits a high degree of fitting with the measured data.The objective function is significantly reduced from 442.11 to 90.44,indicating a descent rate of 79.54%.These results validate the effectiveness and precision of the proposed method.

关 键 词：绕组 梯形网络 参数辨识 反演计算 频响函数 

分 类 号：TM41[电气工程—电器]