Development of a Fault-Tolerant Permanent Magnet Synchronous Motor Using a Machine-Learning Algorithm for a Predictive Maintenance Elevator

Elevators serve as essential vertical transportation systems for both passengers and heavy loads in modern buildings. Electromechanical lifts have become the dominant choice due to their performance advantages over hydraulic systems. A critical component of their drive mechanism is the Permanent Magnet Synchronous Motor (PMSM), which is subject to mechanical and electrical stress during continuous operation. This necessitates advanced monitoring techniques to ensure safety, system reliability, and reduced maintenance costs. In this study, a fault-tolerant PMSM is designed and evaluated through 2D Finite Element Analysis (FEA), optimizing key electromagnetic parameters. The design is validated through experimental testing on a real elevator setup, capturing operational data under various loading conditions. These signals are preprocessed and analyzed using advanced machine-learning techniques, specifically a Random Forest classifier, to distinguish between Normal, Marginal, and Critical states of motor health. The model achieved a classification accuracy of 94%, demonstrating high precision in predictive maintenance capabilities. The results confirm that integrating a fault-tolerant PMSM design with real-time data analytics offers a reliable solution for early fault detection, minimizing downtime and enhancing elevator safety.