Stationary non-Gaussian random vibration control: A review

Study of the response of a single-dof dynamic system under stationary non-Gaussian random loads aimed at fatigue life assessment

“…In addition, an important dependence on damping is observed for both parameters: For very low values of damping, consistent with typical damping ratios of metallic structures in real vibration testing, the response of the system turns out to be Gaussian regardless of the degree of non-Gaussianity of the load to which it is subjected (curves with and ), thus confirming the results already found in the literature; conversely, an increase in the damping coefficient of the system results in a greater degree of non-Gaussianity of the system's response. This result can be explained by considering that the non-Gaussianity of a signal is related to its frequency content in terms of the phase of the Fourier transform [36] . Indeed, the PSD does not provide any information about the Gaussian or non-Gaussian nature of a signal, as it only corresponds to the squared magnitude of the transform, losing information about the phase of the different frequencies.…”

“…Nevertheless, it is crucial to emphasize that the chosen method for load generation was selected in this study because it is the most commonly employed technique in the literature for describing non-Gaussian random loads derived from real-world scenarios. Furthermore, in the case of stationary non-Gaussian random loads, the non-Gaussianity of the signal is linked to the phase of the Fourier transform (rather than its magnitude, as the Gaussianity of a signal is not discernible from its power spectral density) [36] . Therefore, it is reasonable to anticipate analogous results, at least qualitatively (e.g., the influence of system damping and the effect of the Normalized Bandwidth Factor), even for other methods of generating non-Gaussian signals.…”

“…It is worth noting that the corrective coefficients listed above utilize the parameters of skewness and kurtosis regardless of the method used to generate these non-Gaussian values, yielding satisfactory results. Therefore, while it has been noted that skewness and kurtosis parameters alone are insufficient to unambiguously describe a non-Gaussian random signal (i.e., non-Gaussian signals with the same skewness and kurtosis values but different probability distributions can be obtained [36] , [37] ), it can still be asserted that these values represent the two most relevant coefficients to be considered for the calculation of damage induced by a non-Gaussian random load. This assertion holds true irrespective of the specific origin of the non-Gaussianity of such a load [35] .…”

On vehicle response under non-Gaussian road profile excitation

“…4 To simulate the non-Gaussian random vibration caused by road roughness, several methods have been proposed to generate non-Gaussian signals with given PSD and kurtosis. 5,6 Among these methods, the amplitude modulation method was firstly developed by Smallwood,7 in which a Gaussian signal was generated and then multiplied with an amplitude modulation signal (AMS) that was independent of the Gaussian signal. The kurtosis of a non-Gaussian signal generated using this method was governed by the AMS.…”

Random vibration environmental test based on strain response control

“…The spectral parameters calculated from the stress response spectrum, in conjunction with a cycle amplitude probability density function and an appropriate damage accumulation rule, are then utilized to estimate the fatigue life of the structure (Mršnik et al, 2013; Gao et al, 2021b; Luo et al, 2022). Furthermore, the stress response spectrum can be used to generate a stationary Gaussian or non-Gaussian stress time history by supplementing random phase angles (Braccesi et al, 2015; Zheng et al, 2021). Therefore, the stress or strain response spectrum is very crucial for the vibration fatigue analysis.…”