Modeling and control of Diabetes Mellitus via LPV-framework

OData support
Dr. Harmati István
Department of Control Engineering and Information Technology

Because of the developement of technology in the medical field, nowadays it is possible to

perform interventions that can significantly better, or even save the lives of pacients who are

suffering from certain diseases. There are numerous devices and tools that help the doctors

in their work, but there are some physiological processes, where a fully automatized system

can help, because the human organism can not regulate them by itself.

The patients suffering from type I diabetes have to take insulin doses, so they can keep

their blood-glucose concentration on a healthy level. This process can be fully automatized,

the scope of this paper is presenting the design possibilities of a modern, robust controller

for the optimal control of this process. Last semester, during the open-loop simulation

stage and the closed loop phase, where I used a PD controller, the results pointed out, that

the control of the blood-glucose concentration is a diffcult task. The main reason is, that

the modell only represents a small percentage of the patients fully, and the time-varying

dynamic characteristics of the patiens differs between them.

The LPV based control is an effective method used for nonlinear models, the statespace

is constantly depending on one or more parameters that vary in time. In a confined

parameter-area, in every point, the LPV system`s parameters get an exact value, thus

becoming an LTI system. When it comes to robust control, robust-gain-scheduling offers

a solution for such systems, its base priciple is, that the parameter-plain is divided into

smaller areas, and a unique controller ensures the optimal control for that certain area.


Please sign in to download the files of this thesis.