中国电机工程学报
中國電機工程學報
중국전궤공정학보
ZHONGGUO DIANJI GONGCHENG XUEBAO
2015年
10期
2625-2633
,共9页
局部放电%脉冲匹配原子%稀疏表示%匹配追踪%改进量子遗传算法%残差比阈值%去噪
跼部放電%脈遲匹配原子%稀疏錶示%匹配追蹤%改進量子遺傳算法%殘差比閾值%去譟
국부방전%맥충필배원자%희소표시%필배추종%개진양자유전산법%잔차비역치%거조
partial discharge(PD)%pulse matching atom%sparse representation%matching pursuit(MP)%improving quantum genetic algorithm(IQGA)%residual ratio%denoising
局部放电(partial discharge,PD)(简称局放)信号易受噪声干扰,影响监测效果。针对局放信号噪声抑制问题,提出了一种局放信号稀疏表示去噪方法。该方法以信号的稀疏表示及其匹配追踪算法为核心,构建了与局放信号特征相匹配而与噪声信号不相关的局放脉冲匹配原子,并组成过完备原子库。在该原子库中采用匹配追踪算法对染噪局放信号进行稀疏表示,提取最佳局放脉冲匹配原子;并通过改进量子遗传算法加速最佳匹配原子搜索进程,减小计算时间复杂度;同时以残差比阈值作为MP算法迭代终止条件,避免因迭代次数选取不当对去噪结果的影响。最后利用各次迭代提取的最佳脉冲匹配原子仅能对染噪局放信号中原始无噪局放分量进行有效稀疏表示实现去噪目的。运用该方法对仿真信号及实测信号进行了去噪处理,并与基于传统小波理论的局放去噪结果作对比。结果表明,该方法能准确抑制局放信号的噪声干扰,去噪效果优于传统小波方法。
跼部放電(partial discharge,PD)(簡稱跼放)信號易受譟聲榦擾,影響鑑測效果。針對跼放信號譟聲抑製問題,提齣瞭一種跼放信號稀疏錶示去譟方法。該方法以信號的稀疏錶示及其匹配追蹤算法為覈心,構建瞭與跼放信號特徵相匹配而與譟聲信號不相關的跼放脈遲匹配原子,併組成過完備原子庫。在該原子庫中採用匹配追蹤算法對染譟跼放信號進行稀疏錶示,提取最佳跼放脈遲匹配原子;併通過改進量子遺傳算法加速最佳匹配原子搜索進程,減小計算時間複雜度;同時以殘差比閾值作為MP算法迭代終止條件,避免因迭代次數選取不噹對去譟結果的影響。最後利用各次迭代提取的最佳脈遲匹配原子僅能對染譟跼放信號中原始無譟跼放分量進行有效稀疏錶示實現去譟目的。運用該方法對倣真信號及實測信號進行瞭去譟處理,併與基于傳統小波理論的跼放去譟結果作對比。結果錶明,該方法能準確抑製跼放信號的譟聲榦擾,去譟效果優于傳統小波方法。
국부방전(partial discharge,PD)(간칭국방)신호역수조성간우,영향감측효과。침대국방신호조성억제문제,제출료일충국방신호희소표시거조방법。해방법이신호적희소표시급기필배추종산법위핵심,구건료여국방신호특정상필배이여조성신호불상관적국방맥충필배원자,병조성과완비원자고。재해원자고중채용필배추종산법대염조국방신호진행희소표시,제취최가국방맥충필배원자;병통과개진양자유전산법가속최가필배원자수색진정,감소계산시간복잡도;동시이잔차비역치작위MP산법질대종지조건,피면인질대차수선취불당대거조결과적영향。최후이용각차질대제취적최가맥충필배원자부능대염조국방신호중원시무조국방분량진행유효희소표시실현거조목적。운용해방법대방진신호급실측신호진행료거조처리,병여기우전통소파이론적국방거조결과작대비。결과표명,해방법능준학억제국방신호적조성간우,거조효과우우전통소파방법。
Partial discharge (PD) signal is susceptible to noise interference, which will bring influence to monitoring effect. Focusing on the noise suppression, a PD signal denoising method based on sparse representation was given in this paper. The core of this method is signal sparse representation and its matching pursuit(MP) algorithm. The PD pulse matching atom that matching the features of PD pulse signal while mismatching noise signal was designed, and the overcomplete dictionary was built as well. Based upon these, polluted PD signal was sparse decomposed by MP algorithm in this dictionary to search the best PD pulse matching atom. To decrease the calculation time complexity, the searching process of best PD pulse matching atoms were accelerated by improving quantum genetic algorithm (IQGA); morever, to avoid the influence on the denoising result by inappropriate choice of iterations, the residual ratio was chosen to be the terminating condition of the iteration. Finally only the original PD signal could be sparse represented by the best PD pulse matching atoms extracted during iteration and then the goal of denoising was achieved. The denoising method presented in this article was applied on the simulated and fielding measuring signals, the result was critical compared with denoising method based on wavelet as well. The results show that the denoising method of this paper is available to precisely suppress PD signal noise jamming and the denoising effect is superior to the traditional wavelet method.