Comparison between PID-Fuzzy and Numerical Methods Based on Linear Algebra Controllers for Glucose Control in Type 1 Diabetes Treatment

The main problem of treatment in Type 1 Diabetes Mellitus is to maintain glycemia between Normal levels in order to reduce possible long-term nervous and vascular complications. A possible solution for this is an Artificial Pancreas because it automatically provide enough insulin amount during the day. In recent years, a variety of PIDs and MPC controllers have been developed for the proper functioning of the artificial pancreas. This paper presents the design and evaluation of two controllers, a Numerical Methods based on Linear Algebra and a PID-Fuzzy. The simulation is based on 3 meals per day and different parametric variations. The simulation analysis shows that PIDFuzzy maintains blood glucose levels better than the NMLA. The controllers are compared using several performance indexes such as: settling time, ISE and TVu. Finally, a noise source has been added to the glucose sensor in order to see how the controllers are affected.