|
NEURO AND NEURO-FUZZY HIERARCHICAL STRUCTURES COMPARISON IN FDI: CASE STUDY
J. M. F. Calado(1), M. J. G. C. Mendes(1), J. M. G. Sá da Costa(2) and J. Korbicz(3)
(1) IDMEC/ISEL–Instituto Superior de Engenharia de Lisboa
Polytechnic Institute of Lisbon, Mechanical Engineering Studies Centre
Rua Conselheiro Emídio Navarro, 1949-014 Lisboa, Portugal
Fax + 351 21 8317057, e-mail: {jcalado,mmendes}@dem.isel.ipl.pt
(2) Technical University of Lisbon, Instituto Superior Técnico
Dept. of Mechanical Engineering, GCAR/IDMEC
Av. Rovisco Pais, 1049-001 Lisboa, Portugal
Fax + 351 21 8498097, e-mail: sadacosta@dem.ist.utl.pt
(3) Institute of Control and Computation Engineering
Technical University of Zielona Góra
ul. Podgórna 50, 65-246 Zielona Góra, Poland
Fax: + 48 68 3254615, e-mail: J.Korbicz@issi.uz.zgora.pl
In this paper a hierarchical structure of several artificial neural networks has been developed for fault isolation purposes. Two different approaches have been considered. The hierarchical structure is the same for both approaches, but one uses multi-layer feedforward artificial neural networks and the other uses fuzzy neural networks. A result comparison between the two architectures will be presented. It is aimed to isolate multiple simultaneous abrupt and incipient faults from only single abrupt fault symptoms. A continuous binary distillation column has been used as test bed of the current approaches.
Keywords: Fault isolation, neural networks, fuzzy systems, hierarchical structures, fuzzification, white noise.
Session slot T-We-A10: Neuro-Fuzzy Applications of Fault Diagnosis/Area code 7e : Fault Detection, Supervision and Safety of Technical Processes
 | 