The Study Of Processing Track Dynamic-detected Data In Emd-based De-nosing Method And The Software Development Of Smootheness Analysis

At present,track inspection train is applied in cyclical detecting the smoothness of the geometry condition of new-built and operating lines in China.Through the calculation and analysis of dynamic-detected data,it could evaluate the smooth level of tracks,drawing up plans to maintain and guarantee them to run steadily.Owing to the unstable voltage,the abnormal vibration of train body and other factors,dynamic inspection data would be mixed into noise which leads to the inaccuracy of the track quality evaluation.Therefore,it's necessary to slack down the noise in dynamic inspection data before evaluating track quality in order to raise its quality.Empirical Mode Decomposition(EMD for short)was raised for the first time in 1998.Because of its alterable characteristic scale and good adaptivity,EMD has gained attention and been studied gradually.Now,EMD is widely implemented in signal noise reduction,vibration analysis,pectorophony and analysis of seismic waves,etc.In this thesis,we will discuss the theory and the programming of EMD,and apply them to the software development of EVM,which can be used into the noise reduction of dynamic inspection data and evaluating track quality of smoothness.In a word,the thesis mainly includes the three aspects:(1)Studying the theory of EMD and EEMD,comparing them with the theory of wavelet analysis.(2)Studying the threshold denoising methods based on EMD and EEMD,including conventional threshold models of wavelet,the applicative calculation method for EMD theory based on attenuate white noise energy,threshold functions and index of noise reduction.(3)Summarizing three methods of evaluating track quality of smoothness,illustrating the necessity of noise reduction from dynamic inspection data.The thesis studied the theories of EMD and EEMD,two denoising methods and three threshold functions,and analyzed the noise reduced data of gauge,alignment,horizontal offset,height,twist and other inspection items.It compared the noise reduction results of EMD,EEMD and "db1" and studied three evaluating methods,including the score deducting if a part of amplitude is out of gauge,parameters of track quality and power spectrum of irregularity.It compared their advantages and applied range,and took Qing-Rong intercity railways as an example to illustrate its track quality through dynamic inspection data of the railways.According to the content above,the thesis developed a software named EVM,which is based on Visual Studio,Matlab and the theories above.It is able to achieve the functions of reading dynamic inspection data,noise reduction,plotting and evaluating track quality automatically.