粒子滤波在机械密封声发射信号降噪中的应用

Application of Particle Filter in De-noising of Mechanical Seal Acoustic Emission Signal

机械密封声发射信号容易受到环境噪声的干扰,信噪比很低。提出一种基于互补集合经验模态分解(CEEMD)和粒子滤波(PF)相结合的降噪新方法(CEEPF)。该方法首先对采集的声发射信号进行CEEMD分解,利用相关系数原理识别出高频IMF分量后对其进行重构;然后对重构后的信号建立ARIMA模型,将其作为信号的状态方程,再利用小波分解重构思想提取测量噪声,并建立测量方程;最后对重构信号进行粒子滤波,将滤波结果与各低频IMF分量一起重构得到降噪后的声发射信号。结果表明:基于CEEMD与PF的机械密封声发射信号降噪方法能够很好地滤除背景噪声,并且能最大程度保留有效信息。对仿真信号和实验信号分别进行CEEMD小波阈值、标准粒子滤波、CEEPF降噪,发现CEEPF降噪在降噪效果上明显优于其他2种方法。

The acoustic emission signal of the mechanical seal is easily disturbed by the environmental noise,and the signal to noise ratio is relatively low.A new de noising method based on complete ensemble empirical mode decomposition (CEEMD) and particle filter (PF),the CEEPF method was proposed.Firstly,the collected acoustic emission signals was decomposed by the CEEMD,the high frequency IMF component was identified by using the principle of correlation coeffi cient,and then it was reconstructed.Then the ARIMA model was built on the reconstructed signals,which is used as the state equation.Then the measurement noise was extracted using the wavelet decomposition and reconstruction idea,and the measurement equation was established.Finally,particle filtering was performed on the reconstructed signal,and the filtered result was reconstructed with each low frequency IMF component to obtain a noise reduced acoustic emission signal.The results show that the de noising method based on CEEMD and PF can effectively filter out background noise of mechanical seal acoustic emission signals and preserve the effective information to the greatest extent.The simulation signal and the test signal were denoised by CEEMD wavelet threshold,standard particle filter,and CEEPF noise reduction,respectively.It is found that the CEEPF method is obviously superior to other two methods in noise reduction effect.