Robust disturbance observer design with saturations: Application to glucose regulation control for patients with type 1 diabetes

Diatebes mellitus type 1 requires an strict control for insulin delivery to avoid severe consequences derived from hiperglycemia and hipoglycemia. The concept of an artificial pancreas allows the automation of the treatment of patients with this disease. However, capable control algorithms are required to operate efficiently such devices in order to maintain the blood glucose concentration in appropriate levels (GC). Such GC levels together with the incapability of controller to eliminate insulin from system indicates that input and output are bounded, which is taken into account in the controller design to provide a better performance. Owing to the presence of uncertainty and external perturbations, the design of a robust bounded controller based on an Enhanced Generalized Extended State Observer (EGESO) is proposed to ensure an appropriate operation that avoids such scenarios. Estimation of states and disturbance is provided by EGESO, which eliminates the necessity of information related to meal times and carbohydrate content. Closed loop system stability with the proposed controller is guaranteed by means of the Attractive Ellipsoid Method and the solution of a BMI optimization problem. The performance of the proposal is verified by simulations in Simulink, which shows that the proposal emulates the open-loop therapy where the patient delivers a bolus of insulin in parallel form with every meal.