Návrh nového algoritmu SOMA pro optimalizaci s omezeními

Abstract

The thesis is focused on the modern swarm algorithms and constrained optimization. The goal of this thesis is to improve the original version of the Self-organizing migrating algorithm (SOMA). The gradual development leading to finding a suitable solution is described in this thesis. Every partial solution is commented and compared with the original SOMA strategies. The novelty of the developed algorithm lies in the principle of migration of individuals - the individuals move in the direction of two adaptively shifting target locations. The method of perturbation was modified too. In the present algorithm, the generators of chaotic numbers are utilized. The novel algorithm described in this thesis was compared with selected state-of-art algorithms. As the testing problems Pressure vessel design problem, Welded beam design problem, Tension/Compression Spring design problem and functions from the CEC 2017 benchmark were selected. Based on the experimental results, we can conclude that the innovations implemented to the original algorithm of SOMA significantly improved its performance.