A Novel Method for Simulating Transport Vibrations
The paper describes a novel technique for synthesizing nonstationary, non-Gaussian randomvibrations in the laboratory for the physical simulation of transportation hazards. While many naturallyoccurringand artificially produced random processes can be characterised by the Gaussian distribution,wheeled vehicle vibration generally cannot. The proposed technique for synthesizing non-Gaussianvibrations makes use of the fact that wheeled vehicle vibrations have been shown to be nonstationary and toconsist of a sequence of Gaussian segments. This is usually manifested by an elevated overall kurtosis. Thenonstationary (hence non-Gaussian) effect is achieved by combining the rms distribution with the spectralfunction (PSD) of the vibrations. The paper describes how these stochastic fluctuations in rms level areproduced in accordance with a statistical model describing the segment length distribution, itself obtainedfrom vibration data recorded from a broad range of vehicle and route types. Synthesis is achieved by firstcomputing a modulation function which is produced from the rms and the segment length distributionfunctions both of which were developed in previous research. This is achieved by firstly generating asequence of uniformly-distributed random numbers scaled to the range of segment length which itself is afunction of the desired total duration of the synthesized process. In order to transform a uniformly-distributedrandom variable into any arbitrary non-uniform distribution, the cumulative distribution function isestablished and used as a transfer function applied to the uniformly-distributed random variable. Thismodulation function is applied to a Gaussian random signal itself generated by a standard laboratory randomvibration controller (RVC) by means of a purposed-designed variable gain amplifier system. In order tocounteract the feedback function of the RVC, a second variable gain amplifier is introduced into the systemin order to attenuate the feedback signal in inverse proportion to the gain applied to the command signal.This results in a nonstationary, non-Gaussian random signal that statistically conforms to the desired PSD aswell as the RMS distribution function. Results from experiments using a variety of spectral functions andstatistical distribution of the rms show that the technique can be successfully implemented to accuratelysynthesize nonstationary, non-Gaussian vibrations that are typical of vibrations produced by vehiclestravelling over pavements of varying roughness at arbitrary speeds.
M.A.Sek V.Rouillard
School of Engineering and Science Victoria University Melbourne, Australia
国际会议
The 17th IAPRI World Conference on Packaging(第17届世界包装大会)
天津
英文
528-536
2010-10-12（万方平台首次上网日期，不代表论文的发表时间）