Detección y diagnóstico de fallas utilizando la transformada wavelet

Abstract

This paper presents an approach to analyze different kinds of chemical process instrumentation faults. The proposed approach makes a detection and diagnosis of the instrument worsening in the process. To differentiate valve and transmitter problems an analysis in the wavelet domain is presented along with pattern recognition procedures, such as correlation coefficient and Euclidean distance approaches.