Research Of Train Bearing Wayside Fault Diagnosis Under Complex Acoustic Environment

Our research focuses on solving main problems of strong noise, steep distortion and multi-source resulted from wayside train bearings acoustic condition monitoring and fault diagnosis, and the research object is selected as the NJ(P)3226X1bearing which plays an active role in China Railway High-speed (CRH). In view of the analysis difficulty caused by nonlinearity, nonstationarity and coupling of the recorded acoutic signal, many advanced algorithms, such as Doppler correction, re-sampling method in time domain, multi-source separation and paprameters matching, are proposed and introduced to deal with the challenge, at the same time the merits of these methods are evaluated, which lays foundations and provides solutions for the development of wayside train bearings acoustic condition monitoring and fault diagnosis system.First, the main structure parameters of NJ(P)3226X1bearing are presented. Taking the characteristics of wayside acoustic signal into consideration, we propose the experiment scheme of static collection and dynamic transmission, where fault bearing acoustic signal was first collected at the static experiment situation and then set as the sound source surrounded by other noisy sources on a moving vehicle which passes by a microphone with a constant speed to acquire the wayside acoustic signal corrupted by the Doppler Effect. Three kinds of acoustic sources are employed in experiments, and the scheme and significance of each experiment are presented, which enables the reseach to be proceeded successfully.Afterwards, the fault diagnosis method for wayside train bearings acoustic signal under Doppler Effect is explored, where the signal transmission process and the principle of the Doppler distortion are elaborated. What’s more, the Doppler distortion model for wayside acoutic signal is established in accordance with the kinematics and acoustics theory. The proposed wayside train bearings fault diagnosis method of Doppler distortion correction based on instantaneous frequency estimation and time domain re-sampling as well as the defective bearing detection based on improved Dopplerlet transform and Doppler transient matching reveal the influence of Doppler Effect on wayside train bearings acoustic condition monitoring and fault diagnosis system from two aspects:Doppler Effect removal and Doppler Effect embedding. Moreover, these two methods are both free of pre-measurement, which enhances the practicability.Finally, a detailed discussion and analysis concerns the noise location, variety and position relationship between the noise source and target bearing is conducted to dispose the strong noise and multi-source problems of wayside acoustic signal. A nonlinear de-noising method based on variable digital filtering is proposed for multi-source separation and information extraction of the wayside acoustic signal, which could filter out the useful signal and locate the target acoustic source simutaneously so that this method could effectively suppress noise and extract feature information for fault diagnosis. The simulation and experiment both demonstrate the effectiveness of de-noise and multi-source separation of this method.