Invertibility and Neural Networks based FDI Filter

A. Ríos-Bolívar, F. Rivas-Echeverría, and G. Mousalli-Kayat (Venezuela)


Fault detection and Isolation, System Invertibility, Neural Networks, Fault Detection Condition.


In this paper, a method for designing an invertibility and Neural Networks-based fault detection and isolation filter for dynamical systems is presented. The technique consists of establishing an invertibility-based fault detectability condition for the diagnostic model, which is characterized using the faults (inputs) and the system outputs. If the system is invertible, then the fault detection and isolation can be obtained via the reconstruction of the fault modes. The fault modes reconstruction is achieved using a neural networks-based inverse model.

Important Links:

Go Back