Prediction of forming limit diagrams from tensile tests of automotive grade steels by a machine learning approach

Steels’ formability is extremely important to the automotive industry. This property is usually assessed through the Forming Limit Diagrams (FLDs), which comprise the regions of maximum strains, where neither necking nor fracture of the material is observed. In this work, the use of Machine Learning algorithms was evaluated for the FLDs’ construction, based on the tensile properties of a group of steels. A set of historical data, consisting of tensile tests values and measured FLDs, based on Nakajima method of various AHSS, HSLA and IF steels, was used to build the model. In order to evaluate the predictive capacity of the model, experimental and Keeler-Brazier FLDs were compared. The proposed method was able to predict with great correlation and low mean error the FLDs of those materials. The results show a viable route for predicting FLDs based on tensile properties.