Machine learning methods to predict the fatigue life of selectively laser melted Ti6Al4V components
Alessio Centola, Alberto Ciampaglia, Andrea Tridello, Davide Salvatore Paolino
Abstract
Abstract The aim of the present paper is to predict the fatigue life of Selectively Laser Melted (SLMed) Ti6Al4V components via the process parameters, the thermal treatments, the surface treatments and the stress amplitude, adopting machine learning techniques to reduce the cost of further fatigue testing, and to deliver better predictive fatigue designs. The studies resulted in reliable algorithms capable of predicting trustful fatigue curves. The methods have been trained with experimental data available in the literature and validated on testing sets to assess the extrapolation limits and to compare the different methods. The behavior of the networks has also been mapped by varying one SLM process parameter at the time, highlighting how each one affects the life.