Comparison of Reconstruction Schemes of Multiple SVM's Applied to Fault Classification of a Cage Induction Motor

摘要

Different schemes to reconstruct a multi-class classifier from one-to-one support vector machine (SVM) based classifiers are compared with application to fault diagnostics of a cage induction motor. Power spectrum estimates of circulating currents in parallel branches of the motor are calculated with Welch's method, and SVM's are trained to distinguish health spectrum from faulty spectra and faulty spectra from each other. Majority voting, a mixture matrix and neural network are compared in reconstructing the global classification decision from outputs of SVM's.