| In the recent years,the high-speed railway of China has been developing rapidly,and it has completed 25 thousand kilometers so far,ranking first in the world.However,with the gradual improvement of the operation speed of the train,some of the orbital structures have been found to be slightly damaged under the impelling loads of train,and which lead to the destruction of the structure under the rail.This will cause an uneven trajectory,especially the vertical irregularity of the track,the track vertical irregularity is the main factor which causes the vibration of the vehicle and driving speed limiting.Therefore,the change of track geometry must be paid attention to in the process of operation of high-speed railway.Accordingly,the development of the high-speed railway in our country has been turned gradually from the mass construction to the management and maintenance of long-term operation,and intelligent diagnosis and recognition of high speed railway track geometry has increasingly become the important issue of railway operation and maintenance departments.Therefore,research on the identification of high-speed railway track geometry are carried out as follows:1.The present situation of the high-speed railway development in our country and some engineering problems in the process of operation are briefly introduced,and a brief review is presented concerning on main problems of the high speed railway operation management and maintenance at home and abroad,which pointing out the research priorities and contents of this paper.2.Based on the theory of vehicle-track coupling dynamics and the corresponding vertical uniform model,a vehicle-ballastless track numerical simulation program is written,and gives the related parameters and numerical integration method,and then verify the reliability of the program.3.The basic principle and network design process method of neural network are briefly introduced,and the time series prediction and pattern recognition analysis tools based on MATLAB neural network toolbox are introduced.Design process of BP neural network and radial basis network are introduced emphasically.Then the identification accuracy and application ability of the two methods are compared.4.Based on the vehicle-ballastless track coupling dynamics model,the dynamic response of vehicle-track system to harmonic irregularity excitation of track is studied,and the variation characteristics of dynamic response with harmonic irregularity wavelength,amplitude and wave number are investigated.On this basis,the distribution characteristics of car body acceleration response index are analyzed with the principle of statistical analysis,and the correlation of the main components for vehicle system dynamic response and harmonic irregularity wavelength and amplitude is also investigated.5.The prediction to identify efficiently of harmonic irregularity wavelength and amplitude is investigated respectively which based on the time series prediction,neural network pattern recognition,BP neural network and RBF network,used dynamic response of vehicle system for the input vectors.At the same time,the recognition rate of different methods is compared.It can be seen from the analysis that the wavelength and amplitude of harmonic irregularity are well predicted by the four recognition forecasting models,and car body acceleration is the main component of the model input vector.The time series is good for long wavelength wavelength,but bad for the short wave length.The pattern recognition tool can give the recognition rate more intuitively,and the recognition rate changes with the input vector quantities.BP neural network has better classification and recognition ability for most samples,but it is difficult to classify the samples of the classification boundary.The RBF network can identify long wavelength well,but bad for the short wave length,and can predict the change trend of recognition amplitude better. |
PDF Full Text Request