ICEEMDAN结合FastICA方法在转子系统降噪提纯中的应用

Noise reduction and purification method of rotor system vibration signal based on ICEEMDAN and FastICA

针对转子系统采集得到的非平稳信号中存在着较多噪声,导致分解原信号易出现模态混叠和虚假模态现象,使得降噪提纯效果不理想,特征量无法识别等问题,提出了一种将改进自适应噪声的完备集合经验模态分解(Improved Complete Ensemble EMD with Adaptive Noise,ICEEMDAN)和快速独立成分分析(Fast Independent Component Analysis,FastICA)相结合的转子系统振动信号降噪提纯方法。通过设计实验采集加速度信号进行分析对比后发现,该方法能够有效降低加速度信号中存在的噪声,实验结果中的平均绝对误差(MAE)和均方根误差(RMSE)均有所改善。同时通过设计转子系统轴心轨迹提纯实验,验证了该方法的实用性。

For the non-stationary signal collected by the rotor system,there is a lot of noise,which leads to modal aliasing and false modal phenomena in the decomposed original signal,which makes the noise reduction and purification effect unsatisfactory and the feature quantity cannot be identified.A signal noise reduction and purification method that combines ICEEMDAN and FastICA was proposed.After collecting acceleration signals through design experiments for analysis and comparison,it was found that this method can effectively reduce the noise in the acceleration signal,MAE and RMSE in the results were improved.At the same time,the practicability of the method was verified by designing the rotor system axis trajectory purification experiment.