AI-driven predictive maintenance for gold processing mills: Implementation and experimental evaluation based on a supervised learning framework

This paper presents the development, implementation, and evaluation of an AI-based predictive maintenance (PdM) model designed specifically for gold processing mills. The work builds upon an original model developed using real- world operational data from Zimbabwean gold plants. Unlike previous generic frameworks, this research employs supervised learning algorithms, including logistic regression, decision trees, support vector machines (SVM), and random forests, rigorously evaluated for their effectiveness in detecting equipment failures. The Random Forest classifier demonstrated superior performance with a 98.25% accuracy and strong sensitivity to minority failure classes. A user interface was also developed using Streamlit to facilitate practical deployment and interaction with maintenance teams. The study confirms that AI-based PdM models, when appropriately engineered and evaluated, can significantly reduce unplanned downtime, enhance equipment reliability, and optimize maintenance schedules in gold processing environments.