基于结构风险最小化小波变换的故障信号消噪  

Fault Signals De-noising Based on SRM and Wavelet Transforms

作　　者：胡国胜[1] 

机构地区：[1]广东省科学技术职业学院

出　　处：《电力系统及其自动化学报》2007年第1期39-41,95,共4页Proceedings of the CSU-EPSA

基　　金：国家自然科学基金项目(50077008);广东省自然科学基金项目(033044)

摘　　要：基于结构风险最小化方法将改进的小波变换用于故障信号消噪,它考虑的三个重要因素是:基函数,基函数排序和基函数个数选取。通过对小波变换多尺度分解后的系数与频率相除所得的值进行排序,再根据其值的大小取舍,有效地抑制了高频小波系数,避免了传统的基于经验风险最小化理论的小波变换方法在信号处理中过于依赖小波基函数的不足,显示了基于结构风险最小化方法的合理性。通过仿真计算和电机故障信号消噪表明,新的方法在电机故障信号消噪中比一般小波方法效果好,精度提高1倍以上。In this paper, a new method of signal de-noising is presented with three important factors being taken into consideration,that are mother functions, function order and the number of functions, and the proposed method is based on structural risk minimization （SRM） and wavelet threshold method. After sorting the values which are gained through the way of wavelet transform multiresolving coefficients dividing their frequencies, the larger values are reserved while the smaller ones are discarded, which can effectively restrain high frequency coefficients as well as avoid the drawbacks of former wavelet methods based on empirical risk minimization （ERM） which depend on choosing mother function excessively. The simulation results and the fault signal de-noising experiment of electrical machine show that the proposed method is better than the former ones with precision improved by 100%.

关 键 词：结构风险最小化 小波阈值 电机 信号消噪 

分 类 号：TM712[电气工程—电力系统及自动化]