An artificial neural network (ANN) model was designed to predict the tensile properties in high-strength, low-carbon bainitic steels with a focus on the fraction of constituents such as PF (polygonal ferrite), AF (acicular ferrite), GB (granular bainite), and BF (bainitic ferrite). The input parameters of the model were the fraction of constituents, while the output parameters of the model were composed of the yield strength, yield-to-tensile ratio, and uniform elongation. The ANN model to predict the tensile properties exhibited a higher accuracy than the multi linear regression (MLR) model. According to the average index of the relative importance for the input parameters, the yield strength, yield-to-tensile ratio, and uniform elongation could be effectively improved by increasing the fraction of AF, bainitic microstructures (AF, GB, and BF), and PF, respectively, in terms of the work hardening and dislocation slip behavior depending on their microstructural characteristics such as grain size and dislocation density. The ANN model is expected to provide a clearer understanding of the complex relationships between constituent fraction and tensile properties in high-strength, low-carbon bainitic steels.
Artificial neural network modelling of cold-crack resistance of high strength low alloy steel 950A
