Comparison of Machine Learning Models for Steel Plate Fault Prediction

Steel plates are significant in many industries. Making steel plates requires careful quality checks and traditional ways to find faults. In the manufacturing sector, machine learning-based steel plate defect detection is important to guarantee excellent product quality and operational success. Faults like cracks, surface abnormalities, or inclusions can be found instantly using automatic detection, lowering the need for manual inspections and speeding up the production process. By analysing historical data from several sensors utilized during the production process, we can train machine-learning models to spot patterns. These models can predict the probability of defects before they appear. This new solution has the potential to change the steel industry by improving efficiency and reducing costs. The study highlights the growing role of machine learning (ML) in automating and enhancing the detection of defects on steel surfaces. Five common machine learning models were assessed in this study: Random Forest (RF), Extra Tree, Logistic Regression, and Light Gradient Boosting Machine Classifier (LGBM). We assessed their performance using metrics like Precision, Recall, and F1 score. The results reveal that the Light Gradient Boosting Machine Classifier (LGBM) is the most effective algorithm.