An Ensemble Voting-Based Framework for Maintenance Decision Support in Mining Centrifuges

Attaining environmental sustainability in industrial operations requires innovative solutions that optimize resource utilization and improve energy efficiency. Central to this initiative is intelligent predictive maintenance powered by AI and ML. Direct support for critical Sustainable Development Goals (SDGs), such as industrial innovation, responsible consumption, and climate action, is derived from mitigating unforeseen equipment failures and enhancing operational efficiency. This research builds on a preliminary investigation in which various ML and DL models—including Deep Neural Networks (DNN), Long Short-Term Memory (LSTM), K-Nearest Neighbors (KNN), Naïve Bayes (NB), Logistic Regression (LR), and XGBoost—were developed to predict equipment failures in industrial centrifuges utilized in mining. This study improves forecasting accuracy and reliability by using an ensemble voting mechanism that demonstrates remarkable performance with an accuracy of 99.80%, a precision of 1.00/0.99, a recall of 1.00/1.00, and an F1-score of 1.00/0.99. These findings emphasize the capacity of ensemble models in predictive maintenance to facilitate early and more precise problem detection, hence optimizing maintenance procedures and reducing operating costs. This study provides a scalable and practical approach to improving operational efficiency and environmental accountability in the mining industry by integrating predictive maintenance with sustainability-focused methods.