ЗАСТОСУВАННЯ ГІБРИДНОГО ЕВОЛЮЦІЙНОГО МЕТОДУ НА ОСНОВІ РОЮ ЧАСТОК ТА МОДЕЛЮВАННЯ ШТУЧНОЇ ІМУННОЇ СИСТЕМИ В ЗАДАЧАХ ОПТИМІЗАЦІЇ

Abstract

A hybrid evolutionary method for solving conditional and unconditional optimization problems in a continuous space based on a swarm of particles and simulation of the HIPSO artificial immune system is considered. Using the method, 30 test problems were solved in a 25-dimensional real space. The results are compared with other known evolutionary methods. It is shown that the method reliably solves 90% of test problems, while in 67% of cases it finds the global optimum faster than competing methods. It is experimentally proven that the proposed method finds the best solution with an error of up to 2.6% on a wide range of real problems with a probability greater than 0.813.