Fractional-Order Adaptive Control Techniques for Artificial Pancreas

Abstract

Type 1 diabetes mellitus is a disease where the patient is not able to produce necessary insulin to regulate the concentration of glucose in the blood. Artificial pancreas is a device that can regulate this concentration and turn the behavior to normal. The human regulatory system can be modeled using differential equations; their order could be integer or fractional. In this work, we examine the accuracy of fractional-order modeling of the minimal model using real data, then robust control techniques are implemented. First, a model reference indirect adaptive controller is designed using two approaches: integer order approach and fractional-order approach, then a fractional-order sliding mode controller is implemented with a robust sliding mode observer to estimate the glucose concentration in the blood. The controller is tuned by a genetic optimization algorithm. Finally, a Neuro fuzzy controller. Several robustness tests are presented (Meal simulation) and evaluated using different types of errors’ criteria.