Interval Type-2 Fuzzy Classifier for Minimization of the Faults Identification Error
Erick Melo Rocha1; Leiliane Borges Cunha1; Ábner César Santos Bezerra1; Walter Barra Jr.1; Carlos Tavares da Costa Jr.1; José Augusto Lima Barreiros1
1 Universidade Federal do Pará
doi:10.20906/CPS/USM-2016-0036
Resumo
The faults diagnosis and detection in real systems is a hard task, mainly due the occurrence of false alarms in the monitoring system teasing unplanned downtime and stoppages in production. To avoid these undesired behaviors some strategies are used to become the monitoring more robust the external interference, with this purpose, in this paper is used an interval type-2 fuzzy system. An interval type-2 fuzzy system do not presented membership function unique or limited, but an uncertainty region, this shaded region known as footprint of uncertainty (FOU), allows considered uncertainties and nonlinearity typical of noisy real systems and frequently associated to the measurement equipments which cause identification errors and consequently can compromise the good function of the monitoring systems, for instance. For the test of the proposed tool will be used parameters of an Auto Regressive eXogenous (ARX) models, and a comparison with classical fuzzy systems is done for quantify and qualify the results.
Palavras-chave: Interval type-2 fuzzy system; Uncertainties analysis; Faults identification techniques.