Machine Learning Prediction for Cementite Precipitation in Austenite of Low-Alloy Steels

This paper presents a machine learning model to predict the γ/(γ + θ) transformation temperature, which is also known as the A_cm temperature in the FeC phase diagram. From the literature, 25, 920 usable data points are collected, and the dataset is analyzed using a boxplot. The hyperparameters of the machine learning models are adjusted using fivefold cross-validation and grid-search techniques. An artificial neural network (ANN) model is selected based on the determination coefficient. The ANN model is compared with an empirical equation to verify the improvement in the accuracy of the model. The significance of the variables was analyzed using the Shapley additive explanations method. Further, the variable prediction mechanisms are discussed.