A proposal method for fault detection and diagnosis in chemical processes instrumentation using wavelet transform

• Marcos Guillén Universidad de Los Andes-Venezuela
• José Paredes Universidad de Los Andes-Venezuela
• Oscar Camacho Universidad de Los Andes-Venezuela

Palabras clave: fault, wavelets, correlation coefficient, euclidean distance

Resumen

This paper presents an approach to analyse different kinds of chemical processes instrumentation faults. The proposed approach makes a detection and diagnosis of the instrument worsening in the process. A chemical process mode! is used to show how the approach works. to differentiate valves and transmitter problems an analysis in the wavelet domain is presented along with pattern recognition procedures. such as correlation coefficient and Euclidean distance approaches.