Fault Diagnosis in the Brushless Direct Current Drive Using Hybrid Machine Learning Models

Abstract

The brushless direct current (BLDC) motor drive is gaining popularity due to its excellent controllability and high efficiency. This paper introduces a fault diagnosis method for open circuit (OC) and short circuit (SC) BLDC motor drives using a hybrid classifier with hybrid optimization. Features such as current, voltage, speed, and torque are considered as the training data. The features are extracted by discrete wavelet transform (DWT) and then employed to train the classifiers to distinguish between fault types and values of response parameters using the support vector machine and Naive Bayes classifier (SVM-NB). To further improve the performance of the system, hybrid chaotic particle swarm optimization (CPSO) algorithms and teaching-learning-based optimization (TLBO) are used. This method is capable of detecting and recognizing the type of faults in the BLDC motor. The developed approach is implemented on the MATLAB/SIMULINK for OC, SC, and no-fault conditions. These hybrid algorithms provide better performance compared to existing approaches with respect to sensitivity, accuracy, and specificity. This improved model achieves about 98.8% accuracy.