Moderní způsoby ovládání energeticky nezávislých měřicích zařízení

Abstract

The aim of this thesis is to find an optimal strategy for the frequency of measerument of energy independent measuring device. This device is located in an enviroment,~where it is not possible to supply energy demands from power grid. The thesis describes particular soft-computing methods that could be used to solve this problem. The final algorithm uses reinforcement learning,~specifically Q-learning. The behavior of the algorithm was tested in the designed simulation. The simulation was programmed in $C\# .NET Core$. The predicted values were calculated using polynomial approximation. Meteorological data over five years and six combinations of learning rate and polonomial degree were used to test the simulation. The testing shows that the quality of the prediction is influenced by the degree of the polynomial.