Stopper clogging is a significant issue that significantly affects the process parameter and quality of steel produced in continuous casting processes. The goal of this work is to predict stopper clogging by concentrating on anomaly detection and clogging prediction. For this purpose, we implement AI, and in particular machine learning algorithms to find trends and abnormalities in the operational data so as to capture temporal and spatial relationships in operational data.

This study aims to evaluate the methodologies based on deep learning neural networks to detect the occurrence of clogging based on historical data of process variables. For the evaluation of model, the performance metrics, such as precision, recall, F1-score, and area under the receiver operating characteristic curve (AUC-ROC), will be used to evaluate the proposed models' effectiveness. The study will help to develop a new algorithm model for real-time prediction and anomaly detection for stopper region of CC Systems.

According to preliminary findings, it is possible to predict the evolution of several parameters, such as casting speed, stopper position and mold parameters that can help to anticipate potential clogging, which therefore enables prompt treatments and preventive measures. The predictive model will show an immense potential for predicting clogging events in the future, which will increase operational effectiveness and decrease downtime. This study tackles a persistent industrial difficulty by combining modern data-driven methodologies to solve the problem which occurs during continuous casting process.