Research On Source Identification Of Vibration And Noise For Diesel Locomotives

This dissertation aims to solve the engineering problem of the noise reduction of a certain type of diesel locomotive in China and mainly focuses on the scientific problem of source identification of vibration and noise.Well-functional signal processing technology and finite element model are indispensable and important parts for realizing the closed-loop system engineering analysis and constructing high-performance digital comprehensive analysis.To achieve this purpose,the author spent six-years effort on developing a source identification signal processing software based on VMD and established a finite element model of structure,acoustic cavity and acoustic vibration coupling of the studied diesel locomotive.Some problems were solved in the process of research including the VMD parameter selection,real BIMF component sifting,the time-frequency resolution divergence and the underdetermined problem of blind source separation.For studying the vibration noise data in the type test,the VMD method was applied on the in-depth research on vehicle structural mode,subsystem vibration characteristics,component vibration characteristics,power room noise characteristics and driver room noise source identification.Finally the problem of noise reduction in driver cab of the studied diesel locomotive was overcome through the intensive research.The following researches were carried out:1.For the VMD parameter setting,the influence of penalty parameter ? and layer parameter K on signal decomposition is studied.The research demonstrates that when the layer parameter K is suitable,the penalty parameter ? is a value related to the signal energy.In order to obtain the optimal decomposition result of VMD,a new method for the selection of penalty parameter ? is proposed and the formula is given.When the penalty parameter ? is determined,the pseudo component will be caused by the change of the layer parameter K.It is shown that the fitting frequency of the pseudo component will change with the change of the layer parameter K,but the real component is independent of the transformation of the layer parameter K.Therefore,a VMD parameter selection method based on BIMF characteristics is proposed.This method can select the true component and eliminate the pseudo component by observing the variation of fitting frequency and fitting damping of BIMF component signal by layer parameter K.2.The linear and nonstable modal analysis method based on VMD steady-state parameters is studied.The simulation results show that this method can not only effectively identify the modal parameters in linear modal test,but also effectively realize the time-frequency characteristics in nonstable modal test.3.In order to analyze the test results reasonably,the finite element analysis of the structural modal of the test vehicle,the finite element analysis of the acoustic cavity mode of the driver's cab and the finite element analysis of acoustic-structural coupling mode are improved.And the theoretical basis for the identification and control of the noise source is established.4.This thesis systematically analyzed the time-frequency characteristics of the vibration source,forced vibration and noise of the diesel locomotive.It is verified that the timefrequency analysis method of VMD steady-state parameters has better resolution than the traditional CWT time-frequency analysis method and HHT method.This method can more effectively reveal the time-frequency characteristics of noise and vibration signals in engineering application.5.In order to improve the adaptability of the VMD parameter selection in the case of insufficient test signal in blind source separation,a data-driven VMD parameter selection method QVMD is proposed.Moreover,based on QVMD,a new method of underdetermined blind source separation using Fast ICA and PCA is proposed.The method can not only process non-stable signals and non-stationary signals,but also realize trace-source separation processing of more source signals with fewer observation signals.The research demonstrates that the main noise sources of the diesel locomotive come from power room reverberation,auxiliary gearbox vibration,diesel engine vibration and vehicle exterior path noise.6.Combined with the conclusion of the analysis method in this thesis and the finite element simulation,the design of noise reduction control for diesel locomotive is carried out.The test results show that the identified noise source features are accurate and the cab noise is reduced by 8.3d BA.This method effectively solves the engineering problem of excessive noise of the diesel engine compartment from a vehicle manufacturer.