Experimental Test of Sliding-Mode Control for Substrate Regulation in a Fedbatch Bioprocess

On the design of a neuro-fuzzy controller — application to the control of a bioreactor

“…The neuro-fuzzy control system's learning consists in the on-line determination of the components of the parameters vectors λ , minimising the criterion (22). Among the different existing methods (Borne et al, 1990, De Larminat andThomas, 1977), one of them is often used: it is one of the so-called gradient methods.…”

“…The previously proposed learning law involves the gradient of the criterion (22). This gradient must be expressed in accordance with the parameters contained in vector λ to implement the proposed control law.…”

“…This kind of process is difficult to control with classical methods (Augusto and Bonomi, 1996, Marsili-Libelli, 1989, Luca et al, 1997, Parulekar and Lim, 1985 due to its high sensibility toward the environment. Indeed, structural parameters of such processes are time variant when external disturbances occur.…”