Overcoming control challenges in the Artificial Pancreas

This study compares recent Control Algorithms of Artificial Pancreas; Proportional-Integral-Derivative (PID) and Model Predictive Control (MPC). PID controller is applicable to many controls problem, often performs satisfactorily and eliminates the offset. However, it can perform poorly in some applications, does not in general provide optimal control and is unstable with integral control. On the other hand, MPC is able to deal with large multivariable constraints and can be used for non-minimal phase. Then this study focus on two issues, the first one is the nonlinearity of the control problem, and it is demonstrated how this can be tackled via asymmetric objective functions, the proposed MPC strategy employs an asymmetric, state-dependent objective function that leads to a nonlinear optimization problem, The second issue is to propose a velocity-weighting mechanism, within an MPC problem's cost function, that facilitates penalizing predicted hyperglycemic blood-glucose excursions based on the predicted blood-glucose levels' rates of change.