In this paper, we combine the advantages of fuzzy logic and neural network
techniques to develop an intelligent control system for processes having co
mplex, unknown and uncertain dynamics. In the proposed scheme, a neural fuz
zy controller (NFC), which is constructed by an equivalent four-layer conne
ctionist network, is adopted as the process feedback controller. With a der
ived learning algorithm, the NFC is able to learn to control a process adap
tively by updating the fuzzy rules and the membership functions. To identif
y the input-output dynamic behavior of an unknown plant and therefore give
a reference signal to the NFC, a shape-tunable neural network with an error
back-propagation algorithm is implemented. As a case study, we implemented
the proposed algorithm to the direct adaptive control of an open-loop unst
able nonlinear CSTR. Some important issues were studied extensively. Simula
tion comparison with a conventional static fuzzy controller was also perfor
med. Extensive simulation results show that the proposed scheme appears to
be a promising approach to the intelligent control of complex and unknown p
lants, which is directly operational and does not require any a priori syst
em information. (C) 1999 Elsevier Science Ltd. All rights reserved.