Edge-AI Digital Twins for Predictive Maintenance in Smart Manufacturing: A Simulated Performance Evaluation

This empirical-style paper presents a simulated study at the intersection of Mechanical Engineering, Industrial AI, and Manufacturing Systems. The study addresses a current research direction in which digital and data-driven methods are being integrated with domain-specific decision systems. 2,160 asset-days and approximately 51,840 hourly equipment records were simulated. The dataset contained 86 seeded degradation events, of which 62 progressed to a modeled intervention threshold. A simulated 180-day discrete-time manufacturing dataset was generated for twelve rotating assets representing motors, pumps, and spindle assemblies. Every asset contained hourly vibration RMS, kurtosis, temperature, current draw, acoustic energy, throughput, and operating-load variables. Synthetic degradation trajectories were created with gradual bearing wear, lubrication loss, and imbalance events. Four maintenance strategies were evaluated: run-to-failure, fixed-interval preventive maintenance, cloud predictive maintenance, and an edge-AI digital-twin strategy. The edge strategy updated a compact gradient-boosting health model every six hours and fused recent sensor observations with a physics-informed degradation state. The edge-AI digital-twin strategy produced the lowest simulated unplanned downtime and the longest median fault- warning horizon. Relative to fixed preventive maintenance, it reduced simulated downtime by 38.6%, lowered avoidable maintenance interventions, and increased modeled production availability from 93.8% to 97.1%. The cloud predictive strategy performed similarly in classification accuracy but showed greater alert latency during high- network-load periods. All numerical observations, experimental assignments, outcomes, and performance values are simulated for methodological illustration; no real participants, institutions, farms, factories, hospitals, or city systems were studied. The results are intended to demonstrate a coherent research design, provide testable hypotheses, and show how a future real-world study could report evidence transparently. The paper concludes with implementation constraints, methodological limitations, and priorities for empirical validation.