Model reference adaptive fuzzy control of blood glucose level in patients with type 1 diabetes with input saturation

Abstract

This study proposes a method that combines adaptive control, model reference, and fuzzy logic with input saturation to control blood glucose levels in patients with type 1 diabetes. The proposed approach was applied to the nonlinear Bergman model, which accounts for meal disturbances and parameter uncertainties. In the adaptive fuzzy control of a reference model, a mathematical representation of the ideal system is used as the model reference, with the adaptive controller attempting to conform the real system’s performance to this ideal model. Furthermore, this method includes an anti-windup compensator to address actuator performance degradation caused by input saturation and system nonlinearities. The controller design was analyzed using Lyapunov stability theory to guarantee the stability and convergence of the parameter estimation error. To assess the effectiveness of the proposed method in tracking the desired blood glucose level and to compare it with other methods, simulations were conducted in MATLAB/Simulink. The results showed that the proposed variable anti-windup adaptive fuzzy controller not only reached the desired glucose level faster and with smoother tracking under disturbances and parameter uncertainties, but also improved the tracking performance (ISE) by approximately 16.3% and 17.8% compared to the controller without anti-windup, and by approximately 18.5% and 24.7% compared to the constant anti-windup approach, for Patient1 and Patient 2, respectively.