基于多级支持向量机的配电网故障识别方法  

Identification Method of Distribution Network Faults Based on Multilevel SVM

作　　者：程慧 陈艳[1] CHENG Hui;CHEN Yan(School of Mechanical and Electrical Engineering,Chizhou University,Chizhou Anhui,247000)

机构地区：[1]池州学院机电工程学院,安徽池州247000

出　　处：《山西大同大学学报（自然科学版）》2024年第3期55-59,共5页Journal of Shanxi Datong University(Natural Science Edition)

基　　金：池州学院校级科研项目[CZ2020ZRZ12]。

摘　　要：构建了母线电压和主变低压侧电流波形的时频矩阵,并应用奇异值分解(SVD)技术提取波形的奇异谱,进而获得表示奇异值大小的奇异谱均值和描述信号复杂度的奇异熵等关键参数作为特征向量。通过仿真和实际测试,验证了该方法在各种典型条件下的识别精度均超过90%,证明了它在识别各种故障类型方面的有效性、适应性和实用性。This research constructs the time-frequency matrix of the bus voltage and the main transformer low-voltage side cur‐rent waveform,and uses SVD technology to extract the singular spectrum of the waveform,and then obtains key parameters such as the singular spectrum mean that represents the singular value size and the singular entropy that describes the signal complexity as a feature vector.Through simulation and actual testing,it was verified that the identification accuracy of this method exceeds 90%un‐der various typical conditions,proving its effectiveness,adaptability and practicability in identifying various fault types.

关 键 词：配电网故障 时频矩阵 奇异值分解 多级支持向量机 

分 类 号：TM773[电气工程—电力系统及自动化]