Fault feature extraction method for urban rail train wheel vibration signals

Abstract

The invention discloses a fault feature extraction method for urban rail train wheel vibration signals. The method comprises the following steps: 1, collecting a wheel vibration signal; 2, performingimproved EEMD filtering on the wheel vibration signal, adding white noise with a mean value of zero into the wheel vibration signal, carrying out EMD decomposition on the denoised signal to obtain a component with instantaneous frequency from high to low, calculating the permutation entropy of the decomposed component, and removing the component with the permutation entropy greater than a specified threshold from the original signal to obtain a filtered wheel vibration signal; 3, performing EMD decomposition on the filtered wheel vibration signals; 4, performing Hilbert transform on the intrinsic mode component to obtain a marginal spectrum of the intrinsic mode component, and completing fault feature extraction of the urban rail train wheel vibration signal. The method has the advantagesof being high in fault feature acquisition accuracy and easy to implement.

Description

technical field [0001] The invention relates to the technical field of urban rail wheel detection, in particular to a fault feature extraction method for urban rail train wheel vibration signals. Background technique [0002] During the operation of urban rail trains, due to frequent stops and starts, and constant friction between the wheels and rails, it is very prone to wheel out-of-round failures, and the overall out-of-roundness of the wheels is difficult to detect manually. It requires professional testing instruments to detect The time limit and the accuracy rate are not high, so it is very important for the safe operation of urban rail trains to study the detection method of the global wheel irregularity feature. [0003] In vibration signal fault detection, the collected signal contains many noises and shock signals. Due to its own limitations, the traditional filtering method cannot achieve the ideal filtering effect, which is not conducive to the extraction of subs...

AI Technical Summary

Problems solved by technology

[0002] During the operation of urban rail trains, due to frequent stops and starts, and constant friction between the wheels and rails, it is very prone to wheel out-of-round failures, and the overall out-of-roundness of the wheels is difficult to detect manually. It requires professional testing instruments to detect The time limit and the accuracy rate are not high. Therefore, it is very important for the safe operation of urban rail trains to study the detection method of the wheel's global irregularity feature.

[0003] In vibration signal fault detection, the collected signal contains many noises and shock signals. Due to its own limitations, the traditional filtering method cannot achieve the ideal filtering effect, which is not conducive to the extraction of subsequent fault signals.

As a classic signal processing algorithm, EMD can decompose the signal from high frequency to low frequency. However, due to the interference of noise such as intermittent signals, the decomposed signal has the problem of modal aliasing.

Images