Early Fault Detection for Induction Motors using ML

This project presents an integrated MATLAB–Machine Learning framework for early fault detection in induction motors, ensuring predictive maintenance and enhanced operational reliability. The system simultaneously detects and classifies six major motor faults broken rotor bars, stator short circuits, ground faults, overloading, eccentricity, and phase voltage imbalance. A MATLAB Simulink dq-model handles overloading, voltage imbalance, and ground faults, while MATLAB’s built-in induction motor module generates datasets for rotor bar, eccentricity, and stator short-circuit faults. Using K-Nearest Neighbors (KNN) and Decision Tree algorithms, the system achieves high accuracy in fault classification, offering a unified, data-driven solution for industrial fault diagnosis and predictive maintenance.

Key Features:

• MATLAB-based motor modeling and dataset generation
• Six fault condition monitoring and diagnosis
• Machine Learning–based fault classification (KNN, Decision Tree)
• Unified framework for predictive maintenance
• Enhanced operational reliability and reduced downtime.

Dataset Link:

https://ieee-dataport.org/open-access/experimental-database-detecting-and-diagnosing-rotor-broken-bar-three-phase-induction