Robust controller for artificial pancreas for patients with type-1 diabetes

Abstract

The target of this paper is to design a simple and an efficient controller for artificial pancreas (AP) system for blood glucose (BG) regulation in type-1 diabetic mellitus (T1DM) patient. Bergman’s intravenous model is chosen for the controller design as the model is minimum ordered model of T1DM patient. A multi-objective output feedback controller has been designed for AP system considering robustness, disturbance rejection, and transient criterion. Considering H∞, pole-placement, and H2 constraints, a control algorithm has been developed and has been solved using linear matrix inequality (LMI) technique. As a testing platform of the intravenous model-based designed controller, UVa/Padova T1DM metabolic simulator has been chosen which uses subcutaneous insulin delivery method. The controller has been tested in the presence of unannounced meal disturbances. Experimental results show that the controller regulates BG level very efficiently with lesser amount of insulin and avoids hypoglycemia effect. In the presence of external noises like glucose sensor noise and insulin pump error, the robustness of the controller has been checked. The performance of the controller has been compared with compound internal model control (IMC) strategy reported earlier for same meal scenario. The designed multi-objective controller gives better performance metrices than compound IMC controller.