Research On Intelligent Decision-making Method For Fault Diagnosis Of High-speed Train Bogie

In the long-term service of high-speed trains,the performance of key components of the train bogie may degrade and fail,which poses a serious threat to the safe operation of the train.During the operation of the train,the bogies are evaluated by installing various types of sensors in different positions of the bogie,which is important for the safe operation of high-speed trains.Through the analysis of the feature extraction of the monitoring data,feature extraction knowledge base is established and the fault diagnosis decision model is constructed.As the support vector machine(SVM)parameter selection has great influence on its performance,the improved particle swarm optimization algorithm is consumed to optimize the selection of parameters.For the bogie 15 kinds of conditions,including the normal status,two kinds of position of the air spring failure,four kinds of horizontal shock absorber failure and eight positions of anti-snake damper failure,a diagnostic framework based on hierarchical strategy is proposed.The specific research work is as follows:1.Through a variety of feature extraction methods for train bogie vibration signals are employed,the knowledge base of vibration signal feature extraction for high-speed train bogie is established,and the fault diagnosis decision model of high-speed train bogie is constructed.And the mathematical model of the diagnosis decision model is described.2.In the fault diagnosis model,the improved particle swarm optimization algorithm is consumed to optimize the penalty factor and the kernel function parameters of the support vector machine.Aiming at the problem that the particle swarm optimization algorithm is easy to fall into the local optimum,an improved method is given.Initially,the speed update formula multiplies the shrink factor.Subsequently,the inertia weight is reduced by the Gaussian function.By testing the published data,the results verify that the improved particle swarm optimization algorithm to optimize the parameters of SVM employed can improve the accuracy of classification.3.Based on the hierarchical strategy,the fault classification framework of train bogie is given,and the classification order of the bogie fault diagnosis of the high-speed train is established according to the actual situation.The classification order is followed by the original car,air spring failure,transverse shock absorber failure and anti-snake damper failure.The original vehicle is identified by using the amplitude of the discrete Fourier transform of the vibration signal.The singular spectrum entropy of the vibration signal,the power spectrum entropy,the wavelet energy spectrum entropy and the wavelet space characteristic spectrum entropy is employed to carry out the failure of the air spring,the failure of the transverse shock absorber and the failure of the anti-snake damper.The fault location of air spring failure is performed by using singular spectrum entropy,power spectrum entropy,wavelet energy spectrum entropy and wavelet space feature spectrum entropy of vibration signal.The improved particle swarm·optimization algorithm is used to optimize the SVM parameters.The results verify that the correctness of the classification results is high,which is higher than that of the existing ones.