-
Rafael Fernando Silva e Souza
Pós-graduação em Engenharia Elétrica e de Computação, Universidade Federal do Rio grande do Norte, RN
-
Mailson Ribeiro Santos
Departamento de Engenharia de Computação e Automação, Universidade Federal do Rio Grande do Norte, RN
-
Luiz Affonso Guedes
Departamento de Engenharia de Computação e Automação, Universidade Federal do Rio Grande do Norte, RN
Keywords:
Electric motor bearings, Decision Tree, Machine Learning, Artificial Intelligence. Palavras-chaves: Rolamentos em motores elétricos, Árvore de Decisão, Aprendizagem de Máquinas, Inteligência Artificial
Abstract
This paper aims to analyze the influence of input variables on the performance of electric motor bearing fault classifiers. For this, the decision tree algorithm is used to determine and order the relevance of each input variable in the classification process. In this case, the relevance of seven time-domain variables associated with motor shaft vibration signals are analyzed (RMS, peak, minimum, maximum, standard deviation, kurtosis and creast factor). Then, a performance analysis is performed between the number of fault classifier inputs by the accuracy obtained by it. With this, one can observe the best relationship between efficiency and effectiveness of the classifier. The results were obtained using data from the benchmark of Case Western University.
