Supervised diagnosis of induction motor faults: A proposed methodology for an improved performance evaluation

In the last years, numerous investigations have been made within the field of faults diagnosis in induction motors. Most of them use data obtained either from the time domain, through advanced techniques in the frequency domain or even by simulation tools. Some researchers have employed a considerable effort in designing sophisticated algorithms to achieve the best performance of the diagnosis system. In this paper, a novel methodology of evaluation is proposed to promote better evaluation practices in the field of diagnosis of induction motors based on supervised classification. A case study is presented where adequate scores are used for evaluating the performance of the proposed methodology, showing meaningful differences when evaluating several classifiers.