Synthesis of severely non-stationary vehicle vibration signals based on Higher-order moments and Hilbert spectrum

Abstract

Vehicle vibration signals can be severely non-stationary under extreme conditions, and the synthesis of such kind of signal is acrucial researchtopic but requires some level of sophistication. This paper reveals how time-varying amplitudes and phase functions affect statistical characteristics of a random signal, and then proposes a novel four-step method to synthesize vehicle vibrations with arbitrary prescribed duration based on Hilbert spectrum and higher-order moments. Numerical simulations have verified that this method can mostly maintain the characteristics of sample data in both time domain and frequency domain, for instance, the probability density function, skewness, kurtosis, marginal Hilbert spectrum, and degree of stationarity. This method can not only provide accurate inputs for Monte Carlo Simulation, but also provide an experimental platform for Laboratory Testing. Further, it could be a reference approach to the simulation of other severely non-stationary scenarios.