Optimized nonlinear robust controller along with model-parameter estimation for blood glucose regulation in type-1 diabetes.

Abstract

Glucose acts as a fundamental energy source for cells and plays a pivotal role in various physiological processes, including metabolism, signaling, and cellular control. Maintaining precise regulation of blood glucose levels is crucial for overall health and equilibrium. To achieve this balance, insulin is administered either orally or through an artificial pancreas (AP) during sleep, utilizing control algorithms based on mathematical models to regulate blood glucose concentration (BGC). The extended Bergman minimal model (EBMM) is an advanced mathematical framework that incorporates a state variable to accommodate disturbances in insulin levels triggered by factors such as meal intake or exercise-induced sugar burning. In our study, we propose robust nonlinear controllers: adaptive backstepping sliding mode control (AB-SMC), and adaptive backstepping integral super twisting sliding mode control (ABIST-SMC) and compare these with the backstepping sliding mode control (B-SMC) for stabilizing BGC in type 1 diabetic patients. These controllers aim to regulate blood glucose levels in type 1 diabetic patients by providing robust and adaptive control strategies that mitigate disturbances and ensure stability, ultimately enhancing health outcomes and quality of life. Moreover, adaptive parameter estimation is incorporated to eliminate the need for exact model parameter values for control design. The controller gains are meticulously fine-tuned using improved grey wolf optimization, with the integral time absolute error serving as the objective function. Notably, the ABIST-SMC controller emerges as the most efficient, achieving the desired reduction level in less than 1.92 min. The stability of the proposed controllers is rigorously analyzed using the Lyapunov control theory, demonstrating their capability to achieve asymptotic stability. Simulations are conducted to evaluate and compare the performance of the suggested controllers. Additionally, hardware validation is executed using a hardware-in-loop experimental setup.