Vibration Testing by Non-gaussian Random Excitations with Specified Kurtosis. Part II: Numerical and Experimental Results

This paper is a follow-up to a preceding paper (Part I) in which two methods of non-Gaussian random vibration testing with adjustable kurtosis were introduced and motivation for kurtosis control as a way of increasing or decreasing the excitation crest factor was discussed. The current paper (Part II) adds numerical examples of automobile vibration simulation and experimental results for a kurtosis upgrade implemented in the same form of closed-loop control as in industrial shaker controllers. It was observed in experiments that the dynamic range of a kurtosis controller based on the polynomial transformation method was reduced and the handling of resonances worsened notably. These problems also arise with the sigma clipping technique of crest factor limiting. However, there are no such difficulties with the non-Gaussian method of phase manipulation in the inverse fast Fourier transform (IFFT). When using this method, the signal-to-noise ratio, the controller’s dynamic range, and the stabilization time are as good as in standard Gaussian random testing. Evaluation of the performance of the proposed phase selection algorithm has shown that for increased kurtosis it ensures realistic variability of high peaks in terms of their amplitudes and positions, as well as the number of severe peaks per data block. Because of the analytical solution advantage, both methods, the polynomial transformation and the phase selection, meet time restrictions critical for the operation of shaker testing controllers.