On the Study of Induction Motor Fault Identification using Support Vector Machine Algorithms

Induction motors are ever-present in various commercial and industrial processes owing to their high-power capacity, dependability, and low manufacturing costs. However, during their operation their susceptible to heat, mechanical stress, electrical stress, and corrosion. These stresses do not affect the operation of induction motor, however, in a long run it may develop into a major fault which can cause additional maintenance costs and unscheduled downtime, resulting in overall production loss, high financial loss, and sometimes serious human injuries. This work investigates the performance of an induction motor based on various faults conditions. A MATLAB/Simulink platform is used to develop an induction motor model. The performance of induction motor is evaluated on the three-phase line connected to it since implementing a fault in a line affects the performance induction motor. The faults that were tested were no fault condition and faults between AB, AC, BC, AG, CG, BG as well as ABCG. Training dataset was extracted from the sinusoidal waveforms of this various fault conditions. The data is given to the classification learner app in MATLAB for learning. Furthermore, this data set imported to workspace for the training purposes in the classification learner app. Different SVM learning algorithms are trained to deduce the highest learning algorithm in prediction time. The highest SVM learning algorithm according to the results is cubic and Fine Gaussian SVM learning algorithms which shows 100% accuracy on prediction time. The confusion matrix is then used to compare the true predicted and the false predicted classes.