Unplanned equipment failure remains one of the most persistent causes of production loss in discrete and process manufacturing plants. Conventional time-based preventive maintenance and purely reactive corrective maintenance either waste useful component life or expose the plant to sudden, costly stoppages, because neither strategy responds directly to the actual, observed condition of the equipment. This paper develops and evaluates an integrated predictive maintenance (PdM) framework intended to improve equipment reliability and reduce unplanned production downtime by linking equipment criticality assessment, condition monitoring, feature-level diagnostics, healthindex construction, anomaly detection, remaining-useful-life (RUL) estimation, and maintenance-decision prioritization into a single, traceable methodology. The framework was demonstrated on an 11-asset representative rotating-equipment fleet over a 12-month evaluation period. The bearing-fault classifier achieved a precision of 0.875, a recall of 0.913, and an F1-score of 0.894. Random Forest and an LSTM sequence model were identified as the best-performing classification and RUL-estimation models respectively, the latter achieving an RUL root-mean-square error of 4.6 days. The framework was associated with a 25% increase in Mean Time Between Failures (MTBF), a 25% reduction in Mean Time To Repair (MTTR), a 0.96-percentage-point increase in availability, and a 31.8% reduction in unplanned downtime relative to a preimplementation baseline.

Sonal Bhimrao Sonone, Prof. (Dr.) Shashank S Mishra, "Development of an Effective Predictive Maintenance Framework for Improving Equipment Reliability and Reducing Unplanned Production Downtime" International Journal of Scientific Research and Engineering Development, V9(5): Page(446-451) September - October 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.