基于支持向量机增量学习算法的高压电网短路故障位置自动识别  

Automatic Identification of Short-circuit Fault Location in High Voltage Power Grid Based on Support Vector Machine Incremental Learning Algorithm

作　　者：王洪林 董春林 董俊[3] 李维 高黎明 郭俊 Wang Honglin;Dong Chunlin;Dong Jun;Li Wei;Gao Liming;Guo Jun(Electric Power Research Institute,Yunnan Power Grid Co.,Ltd.,Kunming Yunnan 650011,China;Shandong Kehui Electric Power Automation Co.,Ltd.,Zibo Shandong 255087,China;Electric Power Engineering College,Kunming University of Science and Technology,Kunming Yunnan 650050,China;Yuxi Jiangchuan Power Supply Bureau,Yunnan Power Grid Co.,Ltd.,Yuxi Yunnan 653100,China;Kunming Fumin Power Supply Bureau,Yunnan Power Grid Co.,Ltd.,Kunming Yunnan 650011,China)

机构地区：[1]云南电网有限责任公司电力科学研究院,云南昆明650011 [2]山东科汇电力自动化股份有限公司,山东淄博255087 [3]昆明理工大学电力工程学院,云南昆明650050 [4]云南电网有限责任公司玉溪江川供电局,云南玉溪653100 [5]云南电网有限责任公司昆明富民供电局,云南昆明650011

出　　处：《电气自动化》2022年第4期34-36,共3页Electrical Automation

摘　　要：高压电网环境下,因社会用电量过大易造成严重的短路故障现象。为此,提出基于支持向量机(support vector machine,SVM)增量学习算法的高压电网短路故障位置自动识别方法。按照故障训练特征的确定结果,基于SVM增量学习算法提取线性电网中的不可分边界支持向量,进而处理电网结构中的最小化风险,完成电网故障识别环境搭建。配置高压电网的相量量测单元系数,通过确定故障识别元件,校验短路故障发生的位置,实现高压电网短路故障位置的自动识别。试验结果表明:在传输电量增加的情况下,短路故障电量冗余度极值仅能达到40%;故障位置识别精确性保持在94%。设计方法能够有效实现高压电网短路故障位置自动识别,具有较为优越的性能和一定的应用价值。Under the environment of high voltage(HV)power grid,the excessive social power consumption can easily cause serious short-circuit failures.For this reason,an automatic identification method of short-circuit fault location of HV power grid based on support vector machine(SVM)incremental learning algorithm was proposed.According to the determined results of fault training features,the inseparable boundary support vector in the linear power grid was extracted based on the SVM incremental learning algorithm,and then the minimized risk in the power grid structure was processed to complete the construction of power grid fault identification environment.By configuring the phasor measurement unit(PMU)coefficients of HV power grid,the location of HV power grid short-circuit fault can be automatically identified by determining the fault identification components and verifying the location of short-circuit fault.The experimental results show that the maximum redundancy of short-circuit fault power can only reach 40%when the transmission power increases;the accuracy of fault location identification is kept at 94%.The design method can effectively realize the automatic identification of short-circuit fault location in HV power grid,which has superior performance and certain application value.

关 键 词：增量学习算法 高压电网 短路故障 位置识别 训练特征 支持向量 最小化风险 相量量测单元系数 

分 类 号：TM76[电气工程—电力系统及自动化]