Policy Iteration Algorithm for the Control of Oxygenation

The policy iteration algorithm (PIA) is a quasi non-identifier approach of nonlinear optimal control based on a reinforcement learning and iterative algorithm in order to solve the Hamilton-Jacobi-Bellman (HJB) equation. The synthesized state-feedback controller corresponding to the converged solution should be applicable for the control of cardiopulmonary system. In this article, the simulation results for the control of oxygenation were carried out using a simplified first-order model with time delay based on porcine dynamics. The distinctive results of oxygenation control can then be achieved based on the proposed control strategy. In addition, the practical example of water level for interacting three-tank system, which has the nonlinear dynamics similar to that of the oxygenation, was implemented in order to prove the concept of this control scheme.