Robust Adaptive Neural Control of the Blood Glucose for Type 1 Diabetic Patients in Presence of Meals

In this paper, the blood glucose control for type 1 diabetic patients in the presence of model uncertainties and uncertain meals is considered. In order to present an efficient control approach, it is assumed that the dynamics describe the mechanism of the blood glucose regulation in type 1 diabetic patients are completely unknown. Hence, based on the universal approximation property of the radial basis neural network equipped with an adaptive algorithm, as well as using the minimal learning parameter approach, unknown model dynamics are approximated. Then, based on the feedback-linearization approach and robust adaptive compensation scheme, an appropriate control method is designed to regulate blood glucose in type 1 diabetic patients in the presence of a meal. By a Lyapunov based analysis, it is shown that the closed-loop system of blood glucose control is uniformly ultimately bounded and that the blood glucose of diabetic patients converges to a desired level. Finally, simulation results are shown to demonstrate the efficiency of the designed controller.