This study presents a hybrid maintenance strategy combining Reliability-Centered Maintenance (RCM) and predictive modeling to improve the availability and performance of a cream separator used in a dairy powder production line. The proposed approach follows the standard RCM procedure, including functional analysis, failure mode identification, risk assessment, and prioritization of maintenance tasks. A set of 12 critical failure modes were analyzed based on their effects, consequences, and economic impact. Appropriate preventive and predictive maintenance actions were defined, and a Multi-Layer Perceptron (MLP)- based monitoring model was developed to detect abnormal patterns and estimate the remaining useful life of the machine components. The proposed strategy was tested in a real production environment, resulting in a 40% improvement in equipment availability and a significant reduction in unplanned downtimes and maintenance costs. The results highlight the effectiveness of integrating data-driven techniques into traditional maintenance frameworks for critical industrial equipment.

Keywords