Life prediction for rate-dependent low-cycle fatigue of PA6 polymer considering ratchetting: Semi-empirical model and neural network based approach

Abstract

Based on the experimental results in [1,2], a semi-empirical low-cycle fatigue life prediction model is constructed for a polyamide-6 (PA6) polymer, in which the rate dependence of fatigue life and the effect of ratchetting strain on the life are considered. Since the fatigue damage in PA6 is not only loading-cycle-dependent but also time-dependent, a function to describe the complicate effect of stress rate on the fatigue life is obtained. In the proposed model, the detrimental effect of ratchetting strain on the fatigue life is characterized by introducing a function of mean stress. Comparison of the predicted and experimental results shows that the proposed model presents a good prediction. Furthermore, the neural network based method is also used to correlate the low-cycle fatigue data of PA6. The results show that the established neural network based approach achieves a better prediction to the fatigue life of PA6 with the occurrence of ratchetting than that by the proposed semi-empirical model, which demonstrates a possibility to apply the neural network based machine learning method to deal with the fatigue of polymer.