基于小波-中值降噪的经验模式分解及应用

EMD and Its Application Based on Wavelet-Median Synthesis De-noising

为提高经验模式分解(EMD)质量,降低模式混叠,使分解得到的固有模式函数(IMF)序列能较好地反映设备的振动模式,针对EMD在信号处理方面存在的不足,利用小波消噪对高斯噪声的滤除作用,以及中值滤波对脉冲干扰较好的抑制作用,引入小波-中值组合滤波设计,对待分解数据进行预处理以滤除随机噪声并抑制脉冲干扰。将该方法应用到航空发动机振动信号处理中,能够有效减少各IMF的"筛选"次数和EMD分解层数,降低端点效应在EMD分解过程中的累积作用,抑制模式混叠现象。分解结果表明,使用组合滤波设计能更好地反映出发动机的振动模式和工作状态,并能获取表征高压转子和低压转子振动特征的数据分量。

In order to improve Empirical Mode Decomposition(EMD) quality and decrease mode mixing and reflect the device vibration mode by the resulted intrinsic mode function(IMF) series gained through decomposition,synthesized filter design is introduced.With regard to the EMD deficiencies in terms of signal processing,the synthesized filter above,which makes full use of the wavelet de-noising ability and median filter restraints on pulse disturbance,is used to preprocess the data that is to be decomposed.With the method being applied to aero-engine vibration signal processing,each IMF filtration times and EMD decomposition levels can be effectively reduced,the cumulative function of end effect in the decomposition process can be weakened,and then the mode mixing phenomena will be restrained.The decomposition results show that decomposition with the synthesized filter can better reflect engine vibration mode and operating condition and can obtain the data components that can represent high-pressure rotor and low-pressure rotor vibration characteristics.Therefore,it has laid a foundation for later characteristic sampling and fault diagnosis.