Probabilistic fatigue life prediction model of natural rubber components based on the expanded sample data

Abstract

A novel method for determine the probabilistic distribution model of natural rubber (NR) fatigue life is proposed. Uniaxial tension fatigue tests are carried out on dumbbell-shaped cylindrical specimens. According to the fatigue test data, a support vector machine (SVM) model is established, in which the empirical reliability and the rubber fatigue life are considered as input and output variables, respectively. The SVM model is used to expand the NR fatigue life data. The analysis of the expanded data demonstrates that the NR fatigue life follows a lognormal distribution. Accordingly, the fatigue reliability evaluation of NR components is carried out.