Research On The Key Techniques Of The Train Bearing Wayside Acoustic Fault Diagnosis

Railway transportation plays a very important role in the national economy and the modernization of our country, but the train as a major railway transportation vehicle, the safety problems in the process of its operation directly affect people’s life and property safety and the development of national economy. Train wheel bearing failure is the main cause of the train derailment accident or malfunction, which has attracted extensive attention from home and abroad. Thus the train bearing condition monitoring and fault diagnosis is very necessary and of great significance to the safe operation of the train. Compared with the traditional train vehicle monitoring systems, although the wayside acoustic defective bearing detector system has many advantages, the technological difficulties of acoustic signal processing, such as steep distortion, multiple sound sources, and strong noises still hindering the process of its autonomy and localization.The major research of this paper focuses on the key technical problems of the wayside acoustic fault diagnosis of train bearings, and the Doppler distortion remove method, multiple sound sources separation and strong noise filtering methods and the train bearing acoustic signal fault feature extraction methods are proposed in this paper. And these methods are theoretically elicited and experimentally verified, in order to lay a certain theoretical foundation of the final realization of the homemade wayside acoustic defective bearing detector system. The main research contents are as follows:Firstly the structure, fault characteristic frequency and the main failure forms of rolling bearing were discussed, the mechanism of vibration was elaborated. The analysis of the principle of bearing noise signal and the relationship between noise and vibration was given. An acoustic signal acquisition experimental platform was designed for the active duty passenger train bearing NJ (P)3226X1. The artificial cracks had been set by the wire-electrode cutting machine on the train bearings to obtain the train bearing acoustic signals under different fault conditions, which provide the research object and theoretical basis for resolving the key technical problems of the wayside acoustic fault diagnosis of train bearings.The basic concepts and principles of Doppler distortion were discussed in detail, and the Doppler distortion model of wayside acoustic signal was established form the view of moving acoustics, then a mathematical formula of Doppler distortion was derived. After analyzing the flaws and limitations of the existing Doppler distortion correction methods, a Doppler distortion correction method based on Matching Pursuit Dopplerlet transform(MPDT) and resample was proposed. This method can implement both the Doppler distortion correction of single acoustic source and the Doppler distortion correction of multiple acoustic sources one by one, without knowing the initial motion parameters. Meanwhile, this method was robust to random noise by anti-noise analysis. Finally, based on the analysis and processing of train bearing outer race fault and inner race fault acoustic signals, the experimental results demonstrated the effectiveness and feasibility of this method.A detailed discussion of the current research of signal separation was presented. Then the basic concepts and algorithm of generalized S transform(GST) were given, the details of signal reconstruction method based on the GST and time-frequency filtering(TFF) were discussed, and the corresponding simulation case was also given. As the theoretical method of separating high-speed motion acoustic signal from a single channel signal is extremely vacant in the research at home and abroad, based on the Doppler signal matching features of MPDT, a signal reconstruction method based on MPDT, GST and TFF was proposed to separate high-speed motion acoustic signal, thus achieving the purposes of multiple sound sources separation and strong noise filtering of wayside train bearing acoustic signals.By analyzing unique periodic pulse frequency and side-lobe phenomenon of the rotating machinery status signals, a method of wavelet scale variance slope feature extraction for wayside train bearing acoustic signals feature extraction was proposed, based on the method of wavelet multiscale analysis. Then this method was applied in the analysis of a total of200segments train bearing acoustic signals, which contains normal condition signals, outer race fault signals, inner race fault signals and roller fault signals. Finally, the experimental results showed that, as a train bearing fault feature, the wavelet scale variance slope feature had good properties of clustering and stability in the application of wayside acoustic fault diagnosis of train bearings.