Electronics
Peyman Vafadoost Sabzevar; Ahmad Hajipour; Hamidreza Tavakoli
Abstract
One of the main challenges in the field of control is the use of a stable controller and its lack of dependence on the system model and dynamics so that the input signal is applied to the system based on the existing needs. One of the areas that need to control and apply the input signal is type 1 diabetes, ...
Read More
One of the main challenges in the field of control is the use of a stable controller and its lack of dependence on the system model and dynamics so that the input signal is applied to the system based on the existing needs. One of the areas that need to control and apply the input signal is type 1 diabetes, where people with this disease need constant and regular insulin injections based on blood glucose concentration. Based on this, in this article, two free model methods called the Q-learning algorithm and PID have been used to determine insulin dose, and the results of insulin dose injection show the results and high performance of the Q-learning algorithm in determining insulin dose. This algorithm is one of the methods based on artificial intelligence that discovers the optimal policy based on trial and error. Finally, the Q-learning algorithm has been investigated in the presence of noise and its stability has been proven to ensure the performance of the controller.