High Order Approximate Linearization to Control Glucose Levelin Patients with Type I Diabetes

Please enable javascript in your browser.

Fakhroleslam, Mohammad | 2012

803 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 43271 (06)
University: Sharif University of Technology
Department: Chemical and Petroleum Engineering
Advisor(s): Bozorgmehri Boozarjomehry, Ramin
Abstract:
Three different feedback linearization control methods are applied to nonlinear control of type I diabetes mellitus model. One is exact FBL and the two others are approximate manners based on approximations of the nonlinear system and its duality one-form, utilizing uniform approximation and a homotopy operator, respectively. An extended luenberger observer is also employed to observe the required states of designed compensators. The nonlinear compensators are designed based on a minimal model and applied to the minimal model, as well as Sorensen’s physiological model. In addition to the mentioned close-loop control strategies, a discrete control strategy based on an optimization problem is applied. The purpose is to minimize a defined objective function by optimizing the volume and time of a combined injection of Lispro-NPH or Regular-NPH insulin types. Moreover, four aforementioned methods are applied to twominimal modelswith different sensitivities to injected insulin and are investigated.
Keywords:
Nonlinear Control ; Artificial Neural Network ; Diabetes Mellitus Type 1 ; Discrete Control ; High Order Approximate Linearization ; Exact Linearization

No TOC