APLICACIÓN DE LOS ALGORITMOS EVOLUCIÓN DIFERENCIAL Y COLISIÓN DE PARTÍCULAS AL DIAGNÓSTICO DE FALLOS EN SISTEMAS INDUSTRIALES

Abstract

The fault diagnosis implicates the automatic early detection, isolation and localization of faults that have an effect on industrial systems. This topic is of high interest in order to improve the reliability and safety. This article explores the development of new model based fault diagnosis method. Our proposal is based on the application of the algorithms Differential Evolution and Particle Collision. This approach is illustrated using simulation data of the Two tanks benchmark. With the purpose of analyze the advantage of this approach, principally in the topic of robustness and sensibility, some experiments with noisy data and incipient faults are considered. The results indicate the suitability of this approach