基于回归离散傅里叶级数的车辆轴承多普勒信号校正方法

Vehicle bearing doppler signal correction method based on regression discrete fourier series

由于道旁麦克风与运动车辆之间具有相对运动,导致采集的信号存在多普勒效应,从而增加了轴承声学故障诊断的难度。针对此问题,提出了一种基于回归离散傅里叶级数的车辆轴承多普勒信号校正方法。该方法通过建立车辆运动学模型,得到车辆发声时刻序列和麦克风收声时刻序列;对麦克风采集到的信号进行时间变换,并对经时间变换的信号进行离散回归傅里叶级数非线性拟合;最后通过Tihonov正则化求解得到校正信号频谱图。相比传统校正方法,本文提出的方法可直接获取校正信号的频域信息,在道旁声学轴承故障诊断中有良好的应用前景。

The relative motion between the microphone and the moving vehicle produces doppler effect,which increases the difficulty of bearing acoustic fault diagnosis.To solve this problem,a method of vehicle bearing doppler signal correction based on regression discrete Fourier series is proposed.Firstly,the vehicle sound generating time series and microphone sound receiving time series are obtained through the vehicle kinematics model. Secondly,the signal collected by the microphone is transformed by time,and the time-transformed signal is fitted by discrete regression Fourier series nonlinear fitting. Finally,the corrected signal spectrum is obtained by Tihonov regularization solution.This method can directly obtain the frequency domain information of the correction signal,and can effectively correct the doppler signal.