Thestudyofintelligentsystems,beingabletolearnandtogeneralizepatterns,hasbecomeanareahighlyexploredindiverse fields of science. In the industry area, systems are able to diagnose faults that have been widely studied. The electric motors present fundamental role in industry, since much of the production process depends on its good working. Therefore, avoiding unscheduled stoppage and faults is important in the production process. This paper presents supervised learning approaches to classify three-phase induction motors faults, applying Decision Trees and Random Forest algorithms. The great advantage of using intelligent systems in the motor faults classification is the fact that data collection can be done without interrupting the production process. The input to the proposed classifiers is the audio generated from motor noises, which are obtained using an experimental setup with defective real motors. From Decision Trees structure, we can generate understandable “IF-THEN” linguistic rules, which facilitate the understanding of the results and allow the evaluation between the developed models.
Digital Object Identifier (DOI)
Cristina Flamia de Azevedo, Beatriz; Maria Bressan, Glaucia; Marcos Agulhari, Cristiano; Lucas dos Santos, Herman; and Endo, Wagner
"Three-Phase Induction Motors Faults Classification using Audio Signals and Decision Trees,"
Applied Mathematics & Information Sciences: Vol. 13
, Article 19.
Available at: https://dc.naturalspublishing.com/amis/vol13/iss5/19