A Robust Receding Horizon Control Approach to Artificial Glucose Control for Type 1 Diabetes

The problem of controlling the blood glucose value of a patient suffering from type 1 diabetes is considered. The proposed strategy consists in designing a robust nonlinear model predictive controller based on a minimal nonlinear model. The various uncertainties and disturbances are introduced through the use of a variational model. The control problem is then expressed as a constrained game type minimax optimization problem. The choice of a final cost which ensures good stability properties are detailed. The performances of the controller are exemplified on a virtual testing platform showing its good properties.