University of Twente Student Theses
Exploring the potential of Reinforcement Learning-based adaptive control to the artificial pancreas developed by Inreda Diabetics B.V.
Huggers, T.J.P. (2022) Exploring the potential of Reinforcement Learning-based adaptive control to the artificial pancreas developed by Inreda Diabetics B.V.
Full text not available from this repository.
Full Text Status: | Access to this publication is restricted |
Abstract: | This report challenges conventional methods of tuning a multi-input, single-output control problem with high- frequency disturbances and a large delay in the loop. It suggests an adaptive tuning method based on a Deep Deterministic Policy Gradient Reinforcement Learning strategy. The process is modeled, in order to gain an insight into the problem and allow it to be tuned through a conventional method. The model contains two second-order loops in parallel that act antagonistically towards the output that is in need of control. Transfer functions can be found through linearizing the model, from which plots are made that allow the appropriate type of controller to be selected. The controller is then tuned through the Internal Model Control tuning method. The alternative solution is introduced through theoretical analysis, an mathematical description and the MATLAB implementation. The results show a comparison between the conventional tuning method and the Reinforcement Learning implementation, where the introduced approach shows a better response to the delayed disturbances. It is concluded that the introduced approach could be an improvement to conventional tuning methods as the results prove that adaptive control can be applied to the problem at hand. Because of a small available data-set and limited computing power, it cannot be concluded that the adaptive tuning method also has predictive capabilities. |
Item Type: | Essay (Bachelor) |
Clients: | Inreda, Goor, The Netherlands |
Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
Subject: | 31 mathematics, 42 biology, 50 technical science in general, 53 electrotechnology |
Programme: | Electrical Engineering BSc (56953) |
Link to this item: | https://purl.utwente.nl/essays/92285 |
Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Repository Staff Only: item control page