Research On Reconstruction Method Of Array Signal Detection For High-speed Train Bearing

As important parts of train operation, bearings are prone to erosion and malfunction due to long-term heavy load and other factors. It affects the safety of train operation which makes the fault detection of bearings extremely necessary. Non-contact acoustic diagnosis is one of the most effective diagnostic methods, and related research of it is a hotspot recently. Owing to the detection of acoustic signals based on single microphone is of uncertainty and incompleteness, array detection can be used to improve the signal to noise ratio. How to remove the Doppler effect of array signal detection for high-speed mobile and establish reasonable reconstruction criteria are the key technical issues of engineering application which are urgently need to be solved. To this end,this dissertation focuses on the reconstruction method of array acoustic signal for high-speed train bearings, the specific contents and achievements are as follows:Firstly, we analyzed the mechanism and characteristic frequency of bearing defect and the propagation characteristics of the acoustic signal,and deduced the criteria of the microphone array deployment. The relationship between array length, wheel size, bearing parameters,velocity and other parameters of the relationship between and wheel size are given. The experimental results show that the array length and microphone spacing are independent of train speed. And the deployment criteria laid the foundation for experimental platform construction and algorithm validation.Secondly, for the problems of frequency shift and bandwidth extension of high-speed movement in wayside wheel-bearing acoustic diagnosis, a Doppler Effect removal method is proposed based on TDOA and time-domain interpolation. Through framing, windowing and cross-correlation, the time delay differences among microphones, the moving position and the propagation delay of fault bearings in real time are calculated. The frame iteration method is proposed to increase the location accuracy. Then the receiving time series is calculated to correct the original signal. Simulation results demonstrate that the algorithm can remove the Doppler effect of acoustic signal effectively in low SNR and is suitable for variable speed.For the reason of imperfection and indeterminacy in single microphone receiving signal, a method based on cross-correlation and real-time distance weighting is presented to reconstitute the array signal.Delay difference as judging criteria of overlap between microphones is acquired by correlation analysis. And the distance factor is introduced to reconstitute array signal. Simulation results demonstrate that the algorithm can improve SNR and increase defect characteristic frequency.Finally, a simulation experimental system of high-speed bearings'array detection is designed, which further verify the algorithm effectiveness of Doppler correction and array reconstitution.