Machine learning-based model for fatigue behavior analysis of plasma nitrided AISI 304 steel

Abstract

Surface treatments play critical role in fatigue behavior improvement of metals. In this study, a machine learning-based model was employed to analyze the effects of plasma nitriding as a thermal surface treatment on improving the fatigue behavior of AISI 304 steel. Experimental data, encompassing various plasma nitriding and fatigue loading conditions, were utilized to train different types of artificial neural networks including shallow neural networks and deep neural networks. The inputs to the model were the process parameters of plasma nitriding, including time and temperature, along with the stress amplitude in the fatigue test. The output parameter was the fatigue life. The findings demonstrated that employing deep neural networks led to higher accuracy in the predictions. Furthermore, the obtained results of conducted parametric analyses indicated that the optimal temperature range for achieving the highest fatigue performance lies between approximately 450-550 °C for more than 4 h.