-
Rodrigo D. B. de Araújo
Graduação em Engenharia Mecatrônica, Instituto Federal do Ceará, Campus Fortaleza
-
José S. de Souza Neto
Graduação em Engenharia Mecatrônica, Instituto Federal do Ceará, Campus Fortaleza
-
Thais N. de Carvalho
Graduação em Engenharia Mecatrônica, Instituto Federal do Ceará, Campus Fortaleza
-
Emerson V. A. Dias
Graduação em Engenharia Mecatrônica, Instituto Federal do Ceará, Campus Fortaleza
-
Geraldo L. B. Ramalho
Departamento de Indústria, Instituto Federal do Ceará, Campus Fortaleza
-
Josias G. Batista
Departamento de Indústria, Instituto Federal do Ceará, Campus Fortaleza
Keywords:
Vibration descriptor, machine condition monitoring, IoT, non-invasive system
Abstract
The occurrence of failures in industrial automatic systems is usually related to a change on the machine’s vibration pattern. This work proposes a feature extraction based on the mobile standard deviation and a simple and efficient data classification model to be implemented as local processing of IoT (Internet of Things) systems. The proposed method is evaluated in a simulation of a non-invasive online condition monitoring experiment of a rotating machine, with the objective of detecting the condition of a dry running centrifugal pump. The results obtained reveals that the proposed algorithm has high precision to detect the failure condition on the behavior of the monitored machine. The development of these types of systems makes it possible to integrate them into automated industrial plants.
