Induction Motor Fault Diagnosis Using Support Vector Machine, Neural Networks, and Boosting Methods.

Area of Science:

• Electrical Engineering
• Mechanical Engineering
• Data Science

Background:

• Induction motors are crucial in industrial applications but prone to failures that disrupt operations.
• Accurate and rapid fault diagnosis is essential for maintaining industrial productivity.

Purpose of the Study:

• To develop and validate a robust fault diagnosis technique for induction motors.
• To compare the performance of various machine learning models for motor fault detection.

Main Methods:

• Constructed an induction motor simulator with normal, rotor failure, and bearing failure states.
• Acquired 1240 vibration datasets (1024 samples each) for each state.
• Applied and evaluated Support Vector Machine (SVM), Multilayer Neural Network (MLP), Convolutional Neural Network (CNN), Gradient Boosting Machine (GBM), and XGBoost models.
• Utilized stratified K-fold cross-validation to assess diagnostic accuracy and computational speed.
• Developed a graphical user interface (GUI) for the fault diagnosis system.

Main Results:

• All tested machine learning models demonstrated capability in diagnosing induction motor faults.
• The study verified diagnostic accuracies and calculation speeds of the employed models.
• The developed fault diagnosis technique, integrated with a GUI, proved effective.

Conclusions:

• The proposed fault diagnosis technique shows significant promise for real-world industrial applications.
• Machine learning models, particularly when validated with cross-validation, offer effective solutions for induction motor fault detection.
• The integration of a GUI enhances the usability of the developed fault diagnosis system.