Problem-Oriented Algorithms of Solutions Search Based on the Methods of Swarm Intelligence
Victoria V. Bova, Andrey A. Lezhebokov, Гладков Леонид Анатольевич · 2013
This article offers a modified search architecture, which uses multilevel evolution. It allows paralleling the process of solution search and partially eliminating the problem of preliminary convergence of algorithms. Breaking up the search into stages provides the opportunity to apply different algorithms (the evolutionary, genetic and bee algorithms) at each stage. The mechanisms of evolutionary adaptation directly influence the process of formation of the current population of alternative solutions and creation of a new population of promising solutions based on it. Computational experiments allowed to update theoretical evaluations of time complexity of the suggested algorithms as well as of their efficiency at solving complex practical problems.