Robust Polytopic-LPV Body-Weight-Dependent Control of Blood Glucose in Type-1 Diabetes

Navid Vafamand,Mohammad Hassan Asemani,Saleh Mobayen,Gisela Pujol-Vazquez

doi:10.1109/access.2021.3095216

Abstract

This paper presents a robust H ∞ polytopic Linear Parameter Varying (LPV) controller to regulate the blood glucose level in type-1 diabetes patients. The suggested approach utilizes an observer-based controller and polytopic inexact gain-scheduling scheme to construct the stabilizing injecting insulin resilient against the external meal and snack disturbance in Silico. To assure the closed-loop stability and prescribed H ∞ performance, a Quadratic Lyapunov Function (QLF) is used and the stabilization conditions are derived in terms of Linear Matrix Inequalities (LMIs). The conventional minimal Bergman model is improved by considering the effect of body-weight on the glucose-insulin phenomenon and the new body-weight functions are identified based on real data. Then, for the body-weight-dependent Bergman model, a polytopic-LPV model is obtained. Simulations show the advantages of the developed glucose-insulin model and the robustness of the suggested controller in dealing with the effects of a meal for patients with different weights and avoiding hypoglycemia and hyperglycemia disorders.

Highlights

Diabetes disease is the un-functionality of the glucoseinsulin regulation organic system originated from the ineffective operation of the islets of Langerhans, which releases insulin and glucagon to manipulate the Blood Glucose Concentration (BGC) [1]–[3]
In [24], a controller is designed for the minimal model of the insulin-glucose by means of Linear Parameter Varying (LPV) modeling
Scenario 1: To evaluate the effectiveness of the suggested gainscheduling controller with the gains given in Appendix B, a practical daily meal program is considered in Scenario 1

Summary

Introduction

Diabetes disease is the un-functionality of the glucoseinsulin regulation organic system originated from the ineffective operation of the islets of Langerhans, which releases insulin and glucagon to manipulate the Blood Glucose Concentration (BGC) [1]–[3]. In [5], the Bergman model is considered to design a simple adaptive switching control strategy for the insulin delivery rate. In [21], the high-order sliding mode-based therapy approach is developed for both the Sorensen and Bergman models. By incorporating fuzzy type-2, fractional mathematics, and sliding model observer, the BGC is regulated at 80 mg/dL. To predict the future glucose level, a linearized version of the nonlinear Bergman model is considered. In [24], a controller is designed for the minimal model of the insulin-glucose by means of Linear Parameter Varying (LPV) modeling.

Methods

Results

Conclusion