Research on the Identification of Crack Status Through the Axle Acoustic Emission Signal Based on Local Mean Decomposition and Grey Correlation Analysis

Abstract

Rail transportation plays an important role in the traffic of China, and its security is related to people’s life and property safety. The axle is an important part of rail vehicles, and it is also the key to ensuring railway transportation. In this paper, the main innovation point is a combination of the local mean decomposition (LMD) method and the grey relational analysis method, and a new weighting method based on event importance is proposed. First, the non-stationary signal is decomposed into a pure frequency modulation signal and an envelope signal by LMD decomposition of the acoustic emission signal. The signal is decomposed to the cut-off condition to obtain the component signal containing the global multiple frequencies. The time domain signal of the product function (PF) component is transformed into the frequency domain signal by the Fourier transform. The amplitude spectrum of each PF component signal at different stages is obtained, the root mean square value of the amplitude spectrum is calculated and the characteristic factors of each state are formed as the characteristic parameters of the standard sequence of correlation analysis. On the basis that the effect of feature factors on the event is different in the grey relational analysis, a weight calculation method based on event importance is proposed to improve the grey correlation analysis model. The accuracy of the improved grey relational degree model and the traditional grey relational degree model is verified by using the data on the known crack state, and the accuracy of the risk assessment is calculated. The results show that the accuracy of the weighted grey correlation is higher than that of the traditional grey correlation. The method proposed in this paper is verified by using the axle crack data under different working conditions. The applicability of the proposed model is verified by calculating the crack data under different working conditions, and the evaluation results of the weighted grey correlation degree and the traditional grey correlation degree are compared. The effect of the artificial selection resolution coefficient on the result of risk identification is discussed. Through the curve fitting of the correlation degree, the process of crack initiation and propagation is described, which provides a reference for better understanding of the variation law of crack propagation.