Control of Dynamics Systems Based on Q-learning
Anna Filasova, Juraj Klacik, Jan Kasprisin
DOI:
Abstract
The purpose of the paper is to present an algorithm for solving nonlinear systems control problem based on Q-learning,which is a model-free approach belonging to the adaptive critic family of designs and its advantage over other algorithms of thisfamily is that it does not need a model of the system. The example application is concerning the model of power system.