Application of Adaptive Hybrid Neuro-fuzzy Sliding Mode Control to Bioprocesses

VAN CAUWENBERGHE CECILIA E.; BLANCO ANÍBAL M.; BANDONI J. ALBERTO

Buenos Aires

Congreso; XXII Congreso Interamericano de Ingeniería Química, V Congreso Argentino de Ingeniería Química; 2006

Asociación Argentina de Ingenieros Químicos - Confederación Interamericana de Ingenieros Químicos

Fermentation processes present challenging features for control purposes: high nonlinearity, poorly known interactions and uncertainty; so, they are particularly hard to control. Therefore, classic control strategies render poorly when applied to such systems. Besides the intrinsic complex nature of fermentation processes, an additional complicating issue in their control is the impossibility of on-line measurement of the whole state space (metabolite concentrations). In fact no state variable is available without a considerable delay since samples of the systems have to be withdrawn, quenched and further analysed to ‘measure’ any concentration. Therefore state space observation is usually required. Since conventional controllers do not provide in general both, adequate closed-loop performance and technological constraints satisfaction, alternative strategies should be considered. In this contribution a sliding mode/neural fuzzy controller, which presents attractive features to deal with the aforementioned process characteristics is applied to the Saccharomyces cerevisiae yeast glycolysis. Simulation results show adequate behaviour for set-point tracking, disturbance rejection and plant model mismatch.