Diagnóstico de falhas em ar-condicionado tipo split por meio da análise do caos no sinal da corrente elétrica

Abstract

Due to technological evolution in the last decades, especially in relation to computational techniques for data processing, studies have been expanding for the fault detection and diagnosis (FDD) in equipment in the predictive maintenance field. There is an increasing demand for the installation of artificial air conditioning, mainly for type-split equipment, which represents approximately 72% of the air conditioner units installed in Brazil. Therefore, this research aims to develop a non-invasive method for FDD in split air conditioner equipment through the analysis of the electric current signal, performing an approach through the chaotic variable maximum density (SAC-DM). With the method applied, it was possible to diagnose generated faults in this type of equipment, such as fouling in the evaporator air inlet, fouling in the condenser air inlet, fouling in the air filter and degradation of the compressor capacitor, as well as comparing it with a conventional fault diagnosis method for rotating machines, the fast Fourier transform (FFT). The results showed that with the application of SAC-DM it was possible to obtain an accuracy of 100% in the FDD for single faults, while for simultaneous faults was obtained an accuracy between 96.55% and 100% for detection and between 82.76% and 100% for fault diagnosis, in the scenario evaluated.