中文摘要：

以电力变压器超高频局部放电在线检测系统为例，提出了一种基于复小波变换技术的分离局部放电超高频信号和噪声的方法；研究了三步骤去噪算法，对其最优复小波的选择、复小波分解阶数的确定及其阈值算法的选取作了重点研究。针对不同的噪声给出了不同的阈值算法——惩罚阈值法和SURE法的阈值选取能方便地捕捉噪声范围内存在信号分解的微小的细节信息．而普通阂值法和渐进SURE法的阈值选取能有效地去除噪声；最后，给出了在线去除局部放电超高频信号噪声的步骤和方法。结果表明：利用复小波变换技术．抑制电力变压器超高频在线监测的局部放电信号中的噪声是一种极为有效的优化方法。

英文摘要：

A denoising method based on CWT（Complex Wavelet Transform） is presented for power transformer on- line UHF PD monitoring system. The 3- step denoising algorithm is discussed. The optimal selections of complex mother wavelet,as well as the determination of CWT decomposition level and their threshold algorithm,are emphasized. Different threshold algorithms are applied to different noises：the penalized and SURE threshold algorithms can catch the tiny information merged in noise,while the universal and heuristic SURE threshold algorithms can get rid of noise effectively. The steps and methods of on- line UHF PD signal denoising are introduced. Results show that the proposed method is an effective optimal denoising method for power transformer UHF PD signals.