基于自适应VMD和优化DFNN的剩余电流识别  

Residual current recognition based on adaptive VMD and optimized DFNN

作　　者：张祥珂 王雅静[1] 窦震海[1] 白云鹏 王玮[1] ZHANG Xiangke;WANG Yajing;DOU Zhenhai;BAI Yunpeng;WANG Wei(School of Electrical and Electronic Engineering,Shandong University of Technology,Zibo 255000,Shandong,China)

机构地区：[1]山东理工大学电气与电子工程学院,山东淄博255000

出　　处：《电测与仪表》2025年第3期190-197,共8页Electrical Measurement & Instrumentation

基　　金：国家自然科学基金资助项目(52077221);山东省自然科学基金项目(ZR2020MF124)。

摘　　要：为实现剩余电流装置(residual current device,RCD)快速故障识别,提高用电安全性,提出一种基于自适应变分模态分解(adaptive variational modal decomposition,AVMD)和优化动态模糊神经网络(dynamic fuzzy neu-ral network,DFNN)的故障剩余电流识别方法(AVMD-DFNN)。通过经验模态分解法自适应确定VMD的分解参数,实现剩余电流信号的降噪。提取剩余电流信号的特征参数,经降维处理后作为DFNN识别剩余电流的分类指标。通过最小输出法优化DFNN,去除冗余模糊规则函数,从而实现RCD快速故障识别。仿真结果表明,AVMD-DFNN具有较高的识别准确率和速度,为研制新型自适应剩余电流装置提供了理论参考。In order to realize rapid fault recognition of residual current device(RCD)and improve power safety,a fault residual current recognition method(AVMD-DFNN)based on adaptive variational modal decomposition(AVMD)and optimal dynamic fuzzy neural network(DFNN)is proposed.The decomposition parameters of VMD are determined adaptively by empirical mode decomposition(EMD)to realize the de-noising of the residual current signal.The characteristic parameters of residual current signal are extracted and used as the classification index of DFNN to recognize the type of residual current fault after the dimensionality reduction process.The DFNN is opti-mized by the minimum output method to remove the redundant fuzzy rule functions,so as to realize the rapid fault recognition of RCD.The simulation results show that AVMD-DFNN has high recognition accuracy and speed,which provides a theoretical reference for the development of new adaptive residual current devices.

关 键 词：剩余电流 动态模糊神经网络 变分模态分解 故障识别 

分 类 号：TM93[电气工程—电力电子与电力传动]