Speed and Industrial Induction Motor Fault Control Analysis Using ANN-Digital Twin

This study presents the development and performance evaluation of an integrated Artificial Neural Network (ANN)–Digital Twin system for fault detection and speed control in a 30 kW industrial induction motor. The primary objective is to address the limitations of conventional motor control systems, which often struggle to detect and respond to subtle anomalies in real time, resulting in reduced operational efficiency and unplanned downtimes. The proposed solution leverages a real-time digital twin model enhanced by an ANN to monitor, classify, and correct motor faults effectively. A MATLAB-based simulation framework was used to emulate various motor conditions, including slip variation, torque fluctuations, speed deviations, and vibration disturbances. The ANN model employed sigmoid activation functions and weighted input summations and was trained using synthetic data to minimize prediction error through mean squared error (MSE) loss functions. Fault conditions were injected, and their effects on motor performance were studied. Results revealed that the ANN-Digital Twin system achieved a high classification accuracy of 95.3%, with an MSE of 0.0094, demonstrating excellent fault detection capability. Speed control performance remained stable, achieving 96.5% accuracy, even under simulated faults with ±40 rpm disturbances from the 1480 rpm reference. Control signals scaled proportionally with slip, ensuring timely correction. The reliability index decreased linearly from 0.9 to 0.5 as fault count increased, supporting predictive maintenance strategies. Additionally, the system effectively detected vibration anomalies with an RMS value of 0.49. In conclusion, the ANN-Digital Twin system offers a reliable, intelligent, and scalable solution for real-time fault diagnosis and adaptive control in industrial motor applications.