小波包能量谱和神经网络的开关柜局部放电自动检测方法  

Automatic Detection Method of Partial Discharge in Switchgear Based on Wavelet Packet Energy Spectrum and Neural Network

作　　者：赵昊然 陆智勇 江明 刘立石 ZHAO Haoran;LU Zhiyong;JIANG Ming;LIU Lishi(School of Electrical and Automation,Wuhan University,Wuhan 430000,China;State Grid Lu’an Electric Power Supply Company,Lu’an 237000,China)

机构地区：[1]武汉大学电气与自动化学院,武汉430000 [2]国网六安供电公司,六安237000

出　　处：《自动化与仪表》2024年第2期92-96,共5页Automation & Instrumentation

摘　　要：为可靠掌握开关柜的运行状态,精准检测其局部放电现象,提出小波包能量谱和神经网络的开关柜局部放电自动检测方法。该方法采用数字处理技术和模糊参数识别方法,去除信号中的混频信号,获取新的开关柜运行信号;利用小波包分解方法分解该信号,处理信号中噪声信号的同时,提取局部放电的能量谱特征,将该特征输入径向基神经网络中,完成开关柜局部放电的分类检测。测试结果表明,该方法可有效去除混频信号,并且降低信号噪声;处理后信号的失真率和畸变率极低,应用性能较好;可完成开关柜不同类别的局部放电信号检测。In order to reliably grasp the operating status of switchgear and accurately detect its partial discharge phenomenon,a wavelet packet energy spectrum and neural network based automatic detection method for switchgear partial discharge is proposed.This method adopts digital processing technology and fuzzy parameter recognition method to remove the mixed frequency signal in the signal and obtain a new operating signal of the switchgear.Using wavelet packet decomposition method to decompose the signal,while processing the noise signal in the signal,extract the energy spectrum feature of partial discharge,and input this feature into the radial basis function neural network to complete the classification and detection of partial discharge in the switchgear.The test results show that this method can effectively remove mixing signals and reduce signal noise.The distortion rate and distortion rate of the processed signal are extremely low,and the application performance is good.Can complete partial discharge signal detection for different categories of switchgear.

关 键 词：时频分析 馈线终端 自动化 联调测试 能量谱特征 混频信号 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置] TM933[自动化与计算机技术—控制科学与工程]