基于深度置信网络的三相串联故障电弧检测方法  

Method for Detecting Three-phase Series Arc Fault Based on Deep Belief Network

作　　者：李斌[1] 舒洋 LI Bin;SHU Yang(School of Electrical and Control Engineering,Liaoning Technical University,Huludao 125105,China)

机构地区：[1]辽宁工程技术大学电气与控制工程学院,葫芦岛125105

出　　处：《电力系统及其自动化学报》2023年第7期20-28,共9页Proceedings of the CSU-EPSA

基　　金：国家自然科学基金资助项目(51674136、52104160)。

摘　　要：针对三相串联故障电弧的研究大多只是提供一种能够识别出故障电弧的方法,没有考虑用于工业实时检测的可能性,提出了一种基于深度置信网络的故障电弧检测方法。首先,通过搭建三相异步电机故障电弧实验平台获取不同故障情况下的电流数据,并利用提升小波变换对其进行去噪;其次,通过核主成分分析法KPCA(kernel principal component analysis)提取去噪之后的数据的主成分,减少需要分析的变量;最后,通过PSO优化的DBN网络进行故障识别,与BP神经网络和极限学习机相比,其检测速度更快且准确率达到了98.8%,为应用于实时检测提供了可能性。At present,most of the researches on three-phase series arc fault only provide one method to identify the arc fault,without considering the possibility of industrial real-time detection.Aimed at this problem,an arc fault detection method based on deep belief network(DBN)was proposed.First,the current data in different fault conditions were col⁃lected by establishing a three-phase asynchronous motor arc fault experimental platform and further denoised by lifting wavelet transform.Then,the principal components of denoised data were extracted by the kernel principal component analysis(KPCA)algorithm,so that the variables to be analyzed were reduced.Finally,the fault identification was per⁃formed by the PSO-DBN.Compared with the BP neural network and ELM,the proposed method was faster and reached an accuracy rate of 98.8%,providing a possibility for applications of real-time detection.

关 键 词：故障电弧 故障检测 深度置信网络 提升小波变换 核主成分分析 

分 类 号：TM501.2[电气工程—电器]