Vibration testing of vehicle components by random excitations with increased kurtosis

Abstract

Non–Gaussian random vibration testing with kurtosis control is considered in the paper as a way of introducing realistic high peaks into shaker drive signals, thereby increasing the excitation crest factor. This is required for more accurate simulation of ground vehicle vibrations. Implementing kurtosis, not the crest factor itself, as an additional test specification along with the PSD, leads to closed–form expressions for the non–Gaussian simulation criteria and allows for analytical solutions as a result. An approach of subjecting Gaussian signals to polynomial transformation is simple but has an inherent tendency of introducing frequency distortions jeopardising the PSD simulation. There is no such difficulty with the IFFT phase manipulation approach because the power spectrum is not influenced by the phases. A phase selection procedure capable of modelling random excitations with high kurtosis has been developed and it also works for the opposite case of decreasing kurtosis. Because of the analytical solution advantage, the methods proposed in the paper meet time restrictions critical for closed–loop operation of shaker controllers. Numerical simulations of automobile vibrations with severe time history peaks and experimental work with a complete shaker testing setup were carried out.