A novel stationary non-Gaussian random signal generation algorithm for vibration environmental test

Non-Gaussian random vibration control tests have received extensive attention for replicating real-world strong non-Gaussian random vibration environments. However, there is no unified test standard for non-Gaussian random vibrations. This work proposes a novel stationary non-Gaussian random signal generation algorithm which considers the kurtosis and crest factor simultaneously. A non-Gaussian random signal model is constructed that is composed of the spike component and Gaussian random component. The theoretical first four moments are formulated and the analytical expressions are presented. To obtain the approximate solution of the crest factor, the probability density function and cumulative distribution function of the signal model are derived, and a quantile-based approach is utilized to estimate the maximum absolute amplitude. After that, the frequency domain characteristics of the signal model are analyzed. It is found that the power spectral density of the signal is the power spectral summation of the two components. Based on the theoretical derivations, an algorithm for generating non-Gaussian random vibration is put forward. The theoretical model is verified by the numerical simulation. In addition, a procedure for the generation of the multi-channel correlated stationary non-Gaussian random vibration signals is presented with the proposed model. The simulation results indicate the validity and feasibility of the proposed non-Gaussian signal model, and it is expected to be applied in non-Gaussian random vibration environmental tests.