一种适用于混合三端直流输电线路的故障定位方法  

A Fault Location Method for Hybrid Three-Terminal High Voltage Direct Current Transmission Lines

作　　者：高淑萍[1,2] 杨莉莉 武心宇 周晋宇 宋国兵 GAO Shuping;YANG Lili;WU Xinyu;ZHOU Jinyu;SONG Guobing(College of Electrical and Control Engineering,Xi’an University of Science and Technology,Xi’an 710054,China;Xi’an Key Laboratory of Electrical Equipment Condition Monitoring and Power Supply Security,Xi’an 710054,China;School of Electrical Engineering,Xi’an Jiaotong University,Xi’an 710049,China)

机构地区：[1]西安科技大学电气与控制工程学院,西安710054 [2]西安市电气设备状态监测与供电安全重点实验室,西安710054 [3]西安交通大学电气工程学院,西安710049

出　　处：《西安交通大学学报》2025年第1期37-46,共10页Journal of Xi'an Jiaotong University

基　　金：国家自然科学基金国际合作与交流资助项目(52061635105);陕西省自然科学基金资助项目(2024JC-YBMS-792);国家电网有限公司总部科技资助项目(52094020006U)。

摘　　要：针对因结构复杂导致的混合三端直流输电线路故障定位困难的问题,提出了一种结合变分模态分解算法与改进卷积神经网络(CNN)的故障定位方法(VMD-CNN)。首先,利用PSCAD/EMTDC软件构建混合三端直流输电系统模型,获得故障电流数据,应用克拉克变换对其解耦,获取故障电流的线模分量;其次,对得到的线模分量进行变分模态分解(VMD),得到多个本征模态函数(IMF)分量,选取特征信息最丰富的IMF分量作为VMD-CNN模型的输入;然后,利用高效的分类模型支持向量机(SVM)判别故障发生的区域,将提取到的IMF分量作为SVM输入进行训练学习,可以准确判断出故障发生区域;最后,搭建VMD-CNN模型进行故障定位,挖掘出行波信号中蕴藏的故障信息,同时通过麻雀搜索算法优化CNN中的超参数,实现混合三端直流输电线路的精确定位。仿真结果表明:过渡电阻为100Ω,不同故障位置情况下的定位相对误差均在0.17%以内;故障位置为460 km,不同过渡电阻情况下的定位相对误差均在0.25%以内;过渡电阻为50Ω,不同故障类型情况下的相对误差均在0.3%以内。所提方法能够提升不同故障位置、过渡电阻和故障类型下的定位准确性。To address the challenges posed by complex structures in fault location within hybrid three-terminal high voltage direct current(HVDC)transmission lines,a fault location method based on enhanced convolutional neural network(CNN)is proposed.Firstly,the fault current data of the hybrid three-terminal HVDC transmission system is acquired by modeling the system using PSCAD/EMTDC software,with the fault current being decoupled using the Clarke transform to obtain the line-mode components of the fault current.Secondly,variational mode decomposition(VMD)is applied to decompose the line-mode components into multiple intrinsic mode function(IMF)components,with the most informative IMF component being chosen as input for the VMD-CNN model.Then,an efficient classification model,support vector machine(SVM),is employed to classify the fault occurrence region by training on the extracted IMF components as inputs for SVM,ensuring precise identification of the fault region.Finally,a VMD-CNN model is developed for fault location,extracting fault information from traveling wave signals and optimizing CNN hyperparameters using the sparrow search algorithm to achieve accurate fault location in hybrid three-terminal HVDC transmission lines.The simulation results reveal that with a transition resistance of 100Ω,the relative error in fault location is below 0.17%for various fault locations;at a fault position of 460 km,the relative error is under 0.25%for different transition resistance scenarios;and with a transition resistance of 50Ω,the relative error remains below 0.3%for different fault types.The proposed method enhances fault location accuracy under diverse fault locations,transition resistances,and fault types.

关 键 词：混合三端直流输电 故障定位 变分模态分解 卷积神经网络 麻雀搜索算法 

分 类 号：TM773[电气工程—电力系统及自动化]