Numerical Simulation Of Non-Stationary Random Vehicle Vibration Signals

Vibrations generated by road vehicles are usually random and non-stationary,when vehicles undergo irregular terrain or terrible weather condition,these vibrations even become both non-stationary and non-Gaussian.Comparing with stationary random vibration signals,non-stationary random vibration signals hold time-varying characteristics,and the responses caused by non-stationary random vibrations also feature time-varying and even non-linear.Hence,it is hard to describe and analyze those time-varying and non-linear responses,and it is much more harder to apply dynamical optimization and fatigue analysis into vehicle systems with traditional theories.Because non-stationary random vehicle vibrations exist wildly in natural environments and the research on non-stationary random vehicle vibrations is inadequate,currently the methods to analyze vehicle dynamics are Monto Carlo simulation and vibration testing.However,there are no mature processing,summarizing,and standardizing specifications for non-stationary random data as stationary data;either there are no mature numerical methods to simulate non-stationary random data.In addition,at present most vibration shakers can only generate stationary vibration environments,and the way that stationary random vibrations approximate to non-stationary random vibrations,which is commonly adopted in engineering,could produce over-test and short-test results,and consequently fails to meet the requirements of Environmental Testing and Reliability Testing that simulating environment accurately.Therefore,it is necessary to develop the technology of numerically simulating non-stationary random vehicle vibration signals.Base on summarizing the present status of methods on simulating non-stationary random data and the technique of vehicle vibration testing,concerning on the demands of Monto Carlo simulation and vibration testing of vehicle systems,this dissertation focuses on the topic of numerical simulation of non-stationary random vehicle vibration signals.The main content and conclusions of this dissertation are as follows:(1)Establish a new model to represent non-stationary random vehicle vibration signalsBased on the Hilbert-Huang Transform,this paper builds up two hypotheses and establishes a mathematic model to represent non-stationary random vibration signals.By analyzing the characteristics in time and frequency domain,this model can not only provide the theoretical foundation of describing the random,non-stationary and non-Gaussian peculiarities of vehicle vibration signals,but also reveal the relationship between its Hilbert Spectrum and time-varying amplitudes,and proposes a method to adjust its values of statistical moments through fourth order.(2)Propose a novel method to reconstruct non-stationary random vehicle vibration signalsGrounded on the mathematic model,this paper proposes a novel method to reconstruct non-stationary random vehicle vibration signals.This reconstruction method has been validated by several numerical exemplifications that it can retain not only the same time domain features such as waveform,probability distribution function and statistical moments but also the same frequency domain features such as amplitude-frequency-time distribution,marginal Hilbert spectrum and the degree of stationarity in Hilbert Spectrum as those of sample.This method has the advantages of high accuracy and good maneuverability,and is very suitable to reconstruct non-stationary Gaussian/non-Gaussian random vehicle vibration signals.(3)Propose a novel method to propagate non-stationary random vehicle vibration signalsBased on reconstructing method,this paper proposes a novel method to propagate non-stationary random vehicle vibration signals.This propagation method has been verified by several numerical exemplifications that it can mostly maintain the characteristics of non-stationary Gaussian/non-Gaussian random vehicle vibration samples in time and frequency domains.Moreover,This propagation method is applied into the signal generator of testing platform to generate vibration environments.The laboratory experiments demonstrate that this method could provide more accurate non-stationary Gaussian/non-Gaussian random vibration environments than conventional methods.In summary,supported by National Natural Science Foundation of China,according to the methodology of “theoretical analysis,technique realization,experiment validation”,this dissertation centers on numerical simulation of non-stationary random vehicle vibration signals,and resolve three questions one by one:(1)How to establish a mathematic model to accurately represent non-stationary random vehicle vibration signals?(2)How to build up a numerical simulation method to exactly reconstruct non-stationary random vehicle vibration signals?(3)How to build up a numerical simulation method to precisely propagate non-stationary random vehicle vibration signals? The answers to these questions have not only important theoretical value but also engineering instruction significance for the development of Monto Carlo simulation and laboratory vibration environment testing of vehicle systems.