Influence of Gaussian Signal Distribution Error on Random Vibration Fatigue Calculations

Yuzhu Wang (1) , Roger Serra (1) , Pierre Argoul (2)

1 DivS - Dynamique interactions vibrations Structures
2 IFSTTAR/MAST/SDOA - Sécurité et Durabilité des Ouvrages d'Art

In the study of random vibration problems, Gaussian vibration and non-Gaussian vibrations are usually classified according to the excitation signal. The skewness and kurtosis are usually used to distinguish. Here we discuss a non-strict Gaussian signal, wh ich is the error that exists in skewness and kurtosis and usually unavoidable in actual experiments or signals analysis. Through experiments and simulation calculations, the influence of this error on the traditional fatigue calculation method is discussed . The PSD approach will be discussed primarily, and time domain signals based on the rain-flow counting method will be recorded and verified. Total nine calculation model studied in this process. Finally, through a threshold, the range of skewness and kurt osis is indicated, that within this range, Gaussian signal-based calculations can be continued. By comparing the performance of different methods, a better method for signal adaptability can be obtained.

Domaines: Sciences de l'ingénieur [physics]; Acoustique [physics.class-ph]; Mécanique [physics.med-ph]; Vibrations [physics.class-ph]; Traitement du signal et de l'image [eess.SP]

Yuzhu Wang, Roger Serra, Pierre Argoul. Influence of Gaussian Signal Distribution Error on Random Vibration Fatigue Calculations. Surveillance, Vishno and AVE conferences, INSA-Lyon, Université de Lyon, Jul 2019, Lyon, France. ⟨hal-02188557⟩