A Digital Twin Approach to Smart Monitoring and Fault Diagnosis

Abstract

The continuous monitoring of induction motors, combined with early fault detection, is an essential tool for reducing maintenance costs and preventing unexpected downtime. This paper presents a Digital Twin-based methodology for the real-time diagnosis of incipient inter-turn short-circuit faults in induction motors. This diagnostic strategy capitalizes on the digital twin’s real-time integration of a physical asset and its virtual counterpart. It enables the continuous comparison of measured and simulated currents through residual analysis. The Evolving Similarity-Based Modeling Plus (eSBM+) approach facilitates the identification of the motor’s current operating state by comparing it with previously recorded conditions, supporting the precise detection of the onset and evolution of anomalies. The Reconstruction-Based Contributions Online (RBC-Online) system provides real-time insights into the root cause of the fault. The deviation identification automatically triggers a notification to the responsible technician, ensuring timely and well-informed decision-making. Experimental results validated the effectiveness of the proposed methodology, demonstrating high sensitivity and robustness under varying load conditions.