遗传算法用于局部放电小波自适应阈值去噪  

Adaptive Thresholding Wavelet De-noising for PD Signals Using Genetic Algorithm

作　　者：李剑[1] 程昌奎[1] 江天炎[1] 杜林[1] 王有元[1] 

机构地区：[1]重庆大学输配电装备及系统安全与新技术国家重点实验室,重庆400044

出　　处：《高电压技术》2009年第9期2114-2119,共6页High Voltage Engineering

基　　金：教育部新世纪优秀人才支持计划(NCET-06-0763)~~

摘　　要：小波去噪用于局部放电信号在线监测具有良好的效果,阈值选取与局部放电去噪后信号的畸变具有紧密联系。为提高局部放电监测中小波去噪的自适应能力,并降低去噪信号的畸变率,提出一种小波自适应最优阈值去噪算法,用于变压器局部放电脉冲信号去噪。该方法采用小波对局部放电信号进行分解,在阈值选择时采用基于史坦无偏似然估计(SURE)的最优阈值自适应选择方法,并引入一种新的具有多阶导数的阈值函数,结合二进制遗传算法全局自适应搜索最优阈值,使最优阈值自适应寻优速度大大提高。对局部放电仿真信号和现场局部放电信号的去噪结果表明,该方法与Donoho阈值计算公式及标准软阈值法相比,能更好地去除局部放电信号中的白噪声,去噪信号失真度较小,具有良好的应用价值。Wavelet shrinkage methods are efficient for de-noising of partial discharge （PD） detection. Threshold esti- mation is closely related to distortion of PD signals de-noised by wavelet shrinkage methods. We put forward an a- daptive wavelet threshold algorithm for de-noising of PD signals in order to improve the adaptive property of wavelet de-noising and to reduce distortion of de-noised signal. The wavelet de-nosing algorithm is based on an optimum and adaptive shrinkage scheme. A class of shrinkage functions with continuous derivatives and a genetic algorithm are used for the adaptive shrinkage scheme. The genetic algorithm is helpful to obtaining global optimum thresholds and to reducing much time wasted by the adaptive searching computation. The de noising results of simulative PD sig nals and the field PD signals are presented. The PD signals de-noised by the adaptive wavelet threshold algorithm can remove the white noise effectively and have smaller distortion in waveform than the signals de-noised by using Donoho＇s thresholds estimation and the standard soft shrinkage scheme. This method has good value in practical PD online monitoring.

关 键 词：局部放电 在线监测 小波 去噪 自适应阈值 遗传算法 

分 类 号：TM83[电气工程—高电压与绝缘技术]