IMPROVED ADAPTIVE PREDICTIVE CONTROL WITH A NEW MEAL ANNOUNCEMENT APPROACH

GERMÁN CAMPETELLI; DAVID ZUMOFFEN; MARTA BASUALDO

Buenos Aires

Congreso; 23 Congreso Argentino de Control Automático; 2012

Asociación Argentina de Control Automático

In this paper, an improved adaptive predictive control with robust filter is tested as a new control algorithm to be applied in an artificial pancreas. This time a multiple Finite Impulse Response model approach is used for the system and the perturbation (with meal announcement). The shape of the glycaemia reference trajectory is based on the plasma glucose evolution of a healthy person having thesame meal intake. Finally, a variable weighting factor ($\alpha$) in the cost function is included in order to prevent dangerous glycaemia incursions close to the allowed limits