Fatigue life prediction method based on Rayleigh model modified with spectral width coefficient for arbitrary bandwidth signals

Abstract

The application of the Rayleigh model to the life prediction for broadband random vibration signals tends to be overly conservative. To address this issue, the spectral bandwidth coefficient was incorporated into the Rayleigh model in three different forms to modify the centrality, height, and width of the Rayleigh curve, thereby improving the predictive ability of the model for stress amplitude distribution in the medium-to-high stress range. Additionally, an exponential function was introduced to adjust the probability density distribution in the low-stress amplitude region. The proposed model can accurately reflect the stress amplitude distribution of various bandwidth signals in the low, medium, and high ranges, thereby improving the accuracy of predicting the life of random vibration signals. The vibration fatigue life prediction results for narrowband, non-typical bandwidth, and broadband signals are all within a factor of 2.