Robust guaranteed‐cost output feedback control of blood glucose in type 1 diabetes patient with intrapatient variability

Abstract

SummaryDealing with the interpatient and the intrapatient variability in type 1 diabetes mellitus (T1DM) patient are the prominent control challenges in the development of the artificial pancreas system (APS). An APS consider a nonlinear patient model that has many state variables, but the reality is that out of the many states only glucose concentration is measurable till date without any complication and undue high cost. In the present work, the philosophy is to evolve a control technique that can address the above‐cited challenges or issues. In the sequel, a robust observer is first designed based on the attractive ellipsoid method (AEM) for norm‐bounded uncertainty structure. The information of the estimated states is then used to design a guaranteed‐cost robust output feedback control. The design methodology approximates the nonlinearity present in the model as quasi‐one‐sided Lipschitz thus including more information of nonlinearity in the design as well as the control problem is recast into a linear matrix inequality (LMI) framework. Numerical simulations of 200 virtual T1DM patients are carried out which reveal that there are no cases of hypoglycemia and postprandial hyperglycemia in the presence of bounded parametric variability and high uncertainty in exogenous meal disturbance. The proposed method is simple, efficient and optimal closed‐loop solution for blood glucose regulation in T1DM patients in a robust nonlinear framework.