Electrical Motor Fault Detection System using AI's Random Forest Classifier Technique
Raziq Yaqub, Hassan Ali, Mohd Helmy Bin Abd Wahab
Abstract
Induction motors are very commonly used to create traction for electric vehicles. In electric vehicles and other rotating machine systems, faults develop during regular operation, which, if unaddressed, can accelerate the damage to machines. Therefore, early detection and prediction of faults hold the key importance in predictive and preventive maintenance. This work focuses on collecting sounds and vibrations data of an electric motor, which are the critical indicators of fault initiation in electric vehicle systems. We use artificial intelligence (AI) to predict and classify the types of faults associated with different vibration frequencies and sound levels. We explore random forest classifier technique for faults prediction and derive promising results for different fault conditions. Results show that we correctly predicted and classified different types of manually induced faults associated with different vibration frequencies and sound levels with a 92% accuracy and 86 to 91 % precision level.