Diseño y simulación de un sistema para detección de fallas utilizando el filtro de Kalman como estimador de parámetros para un tanque de mezcla de fluidos en dos configuraciones

Abstract

Fault detection is of grate use in industrial processes, in this work model based fault detection is used, the extended Kalman filter is employed to do parameter estimation in a non-linear plant, the result of the estimation is compared to fault free parameters to generate residues, this are processed whit a neural network to classify faults according to which element they originate from a sensor or an actuator. Training data for the network is generated making simulations of the process.