Robust Backstepping Based Automated Control For type 1 Diabetes

Abstract

Diabetes is a metabolic disorder due to which insulin producing beta cells are destroyed. Therefore, there is less or no production of insulin in the human body and patient has high blood glucose level. To control blood glucose level within the normal range, the idea of artificial pancreas (AP) is used. AP has three parts; a glucose sensor, an insulin pump, and a closed loop control algorithm. In this thesis, the blood glucose-insulin dynamics in human body has been represented by Bergman’s ”minimal model”, having insulin pump dynamics as a state variable to further automate the insulin infusion rate dynamics. In the designing of controllers, the physical nonlinearities, effect of meal disturbance, and sensor noise are also considered. Non-linear techniques like; Sliding Mode Controller plus Back stepping, Integral Backstepping and Backstepping are used to design the corresponding controllers. Asymptotic stability of the proposed controllers has been studied using of Lyapunov theory. For the simulation results Matlab/Simulink environment is used to compare the performance of proposed controllers