Using machine learning algorithms for predicting ultimate tensile strength of steel rebar based on quenching parameters

Constructing a prediction model for industrial processes is beneficial as it can improve the comprehension of systems and processes in industries. Machine learning-based models are vital tools for predicting complex processes. In this work, the efficiency of different machine learning algorithms for evaluating the impact of steel rebar's quenching parameters on the ultimate tensile strength of steel rebar has been examined. For this aim, the following machine learning-based regression algorithms have been implemented: Random Forest, Decision Tree, Support Vector Machine, Artificial Neural Networks, and K-Nearest Neighbors. The models have been trained and tested on ۲۶۹۷ collected data set recorded in Atieh Steel Company, Iran. The performance of these trained models were compared based on different error criteria. The results illustrate that while Random Forest shows the best performance and can estimate the output satisfactorily among different algorithms, the Decision Tree algorithm had the lowest performance in terms of prediction accuracy

Keywords:

machine learning , prediction , quenching , steel rebar , mechanical strength