基于小波包与改进EMD的轮轨力信号降噪

Wheel/Rail Force Signal Denoising Based on Wavelet Packet and Improved EMD

地下列车的轮轨力应变信号在采集过程中,会受到城市上部各种交通车辆、噪声,以及列车车轮磨耗、轨面不平顺等诸多因素的影响。采集的轮轨力信号将失去其准确性。针对轮轨力应变信号中存在的随机白噪声,以及在EMD分解过程中出现模态混叠的问题,提出一种改进的信号处理方法,该方法是一种小波包降噪算法与EMD解相关算法相结合的数据处理方法,能够同时有效抑制模态混叠现象和消除噪声干扰。运用改进的信号处理方法对仿真信号和实测的地下列车轮轨力信号进行处理分析。研究结果表明:该改进的数据处理方法能有效地消除轮轨力在采集过程中随机白噪声的干扰和抑制模态的混叠。对有效地识别轮轨力真实信号具有重要的实际意义。

In the wheel/rail force acquisition process of the underground trains, the strain signal is greatly affected byvarious traffic vehicles on the ground surface, noise, and the train wheel wear, rail surface irregularity and so on. Thus, thecollected wheel/rail force signal may include some disturbance. In view of the random white noise in the wheel/rail forcestrain signal and the problem of mode mixing in the process of EMD decomposition, an improved signal processing methodcombining the wavelet packet denoising with the EMD solution algorithm is proposed. This method can restrain the modemixing phenomenon and eliminate the noise disturbance. Using the improved signal processing method, the simulated signaland the collected signal of the wheel/rail force are analyzed. The results show that the improved data processing method caneffectively eliminate the random white noise generated in the process of wheel/rail force acquisition and suppress the modalmixing. This work has a good practical significance for effective identification of real signals of the wheel/rail forces.