用于复小波变换去噪的复阈值构造算法  

Study of the Construction of Complex Threshold for Complex Wavelet Transform Suppressing Noise

作　　者：唐炬[1] 谢颜斌[1] 周倩[1] 唐铭[1] 彭莉[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：《电工技术学报》2008年第10期121-128,共8页Transactions of China Electrotechnical Society

基　　金：国家自然科学基金资助项目(50577069)。

摘　　要：基于复小波变换的特点,提出分别对其系数的实部和虚部进行独立阈值选取,根据噪声水平用复掩码算子调整阈值,由此得到复阈值的一种新算法。将所构造的复阈值用于一种全新的复合信息——WTRI,对仿真局部放电信号的去噪,并引入波形相似性参数和去噪前后的信噪比,对不同复阈值的去噪效果进行综合评价,由此总结出选择复阈值的一般规律;最后针对不同模拟缺陷类型的实测PD信号,构造了与之对应的最佳复阈值。研究结果表明,采用该复阈值技术的复小波变换抑制噪声的能力更强,而且在高信噪比的情况下,对信号的畸变也更小。According to characteristics of complex wavelet transform, a new threshold algorithm is put forward, which choses threshold of real part and imaginary part of complex wavelet coefficients independently at first, and then, adjusts the threshold by complex mask based on the noise level. In order to prove the validity of a brand-new compound information WTRI, normalized correlation coefficient （NCC） and signal to noise ratio （SNR） for the denoising effect estimation are introduced. In this paper, the optimal complex threshold for different PD signals is constructed and the general rule of choosing the best complex threshold is summarized. Finally, white-noise of real PD signals is eliminated by using complex threshold of complex wavelet. The results show that this method can reduce white-noise effectively.

关 键 词：局部放电 复小波 复闽值 白噪声 

分 类 号：TM835[电气工程—高电压与绝缘技术]