• 2026-05-08
• M.O. Idris, B.S. Adeboye, A.C. Adedayo, A.G Abioye, S.T. Oyewo
• Volume 2, Issue 1

Abstract

This paper formulates and test predictive maintenance models to conveyor belt system in continuous production industries where there are frequent unforeseen downtimes as a result of belt failure, which lead to major economic and operational problems. To solve this, six machine learning algorithms which included Decision Tree Classifier, K-Nearest Neighbors, Logistic Regression, Support Vector Machines, Extra Trees Classifier and Naive Bayes were carried out and contrasted. Industrial fault detection datasets were used to train and test the models and evaluate their performance in terms of accuracy, precision, recall, F1-score and ROC-AUC. The Extra Trees Classifier model displayed the best classification performance (accuracy, precision, recall, F1-score, and ROC-AUC = 1.0) and was the most effective model to identify conveyor belt failures. These results support the idea of shifting the traditional reactive maintenance to the data-driven predictive solutions, which can be aligned with Industry 4.0 goals and address the Sustainable Development Goals 8, 9 and 12, enhancing decent work, industrial innovation, and responsible production. The research concludes that predictive maintenance frameworks involving machine learning, especially those that utilize ensemble techniques, are a sound method of reducing downtime and improving operational efficiency in the industrial setting.