Research On On-rail Sensing Algorithm For Wheel Flat Of High-speed Train

With the increasing of train speed and axle load,the wheel flats occur more frequently.Once the wheel flats appear,the impact force between wheel and rail will be higher dozens or even hundreds of times than the normal condition,and pose a threat to the safety of trains.Hence,it is of great significance to detect the wheel flat and its depth for ensuring the train operation safety and developing the maintenance plan.Firstly,the vehicle-track coupling system model with the track irregularity and the wheel flat is built.Then the rail vibration response is solved by Newmark multistep predictor-corrector algorithm.On this basis,the vibration characteristics are analyzed under some kinds of conditions,such as different track irregularity,different speed and different flat depths.At the same time,the simulation results provide the data for the latter feature extraction.In order to effectively identify the wheel flat and its depth,a pattern recognition algorithm for wheel flats is proposed based on PSO-SVM.In the pattern recognition algorithm,the recognition accuracy will be directly affected by the extracted feature parameters,so a feature extraction method for the rail vibration responses when the train runs on the rail is proposed based on high-order spectrum and gray-gradient co-occurrence matrix.The two-dimensional contour and three-dimensional double spectrum diagrams are obtained by analyzing the rail vibration responses under the condition of normal wheel and wheel flat with high-order spectrum method.Six texture features of the 2D contour map,which are extracted from gray-gradient co-occurrence matrix,are combined with the train speed to detect whether there is a wheel flat.Then the peak value of the diagonal slice of the three-dimensional double spectrum map and the frequency of the inner spread of the two-dimensional contour map are added to recognize different wheel-flat levels.In order to verify the effectiveness of high-order spectrum,Hilbert-Huang transform,EMD decomposition and wavelet packet transform are also used to extract the characteristic parameters of the rail vibration responses.The simulation results show that the detection accuracy of normal wheel and wheel flats can be up to 100%and the recognition accuracy of wheel flat different levels is 95.89%based on high-order spectrum and gray-gradient co-occurrence matrix,better than HHT,EMD and WPT.In order to prove the robustness of high-order spectrum,the applicability of the algorithm under different track irregularities and different wheel defects is also analyzed.The results show that the size of track irregularity will not affect the identification of wheel flat,but will affect the identification of wheel flat levels,and with the increase of track irregularity,the recognition accuracy will be decreased.At the same time,the wheel flat can be completely separated from other defects.The research in this dissertation provides a new solution to detect the wheel flat from the rail vibration responses.