Support vector machines for classification of electrical resistance values within a VSI

Nonintrusive monitoring is a powerful tool to increase the quality, reliability, and efficiency of power electronics systems. In recent years, the use of intelligent systems to perform this type of tasks has been increased due to its high capacity to correctly predict the values of the parameters under monitoring. Support Vector Machines (SVM) is a supervised learning method that maps input data into a high dimensional attribute space through a kernel function to finally find the hyperplane that separates them with the maximum possible margin. In this work, the performance of SVM for estimation of electrical resistance values in a Voltage Source Inverter (VSI) was experimentally proved. Experimental results confirmed the potentialities of this Machine Learning tool for the nonintrusive monitoring of electrical resistance in a VSI.