Home // International Journal On Advances in Systems and Measurements, volume 10, numbers 1 and 2, 2017 // View article

Comparative Analysis of Heuristic Algorithms for Solving Multiextremal Problems

Authors:
Rudolf Neydorf
Ivan Chernogorov
Victor Polyakh
Orkhan Yarakhmedov
Yulia Goncharova
Dean Vucinic

Keywords: searching optimization; multi-extremes; genetic algorithm; swarm algorithm; ant algorithm

Abstract:
In this paper, 3 of the most popular search optimization algorithms are applied to study the multi-extremal problems, which are more extensive and complex than the single-extremal problems. This study has shown that only the heuristic algorithms can provide an effective solution to solve the multiextremal problems. Among the large group of available algorithms, the 3 methods have demonstrated the best performance, which are: (1) particles swarming modelling method, (2) evolutionary-genetic extrema selection and (3) search technique based on the ant colony method. The previous comparison study, where these approaches have been applied to an overall test environment with the multiextremal Rastrigin functions, has shown already their suitability to solve multiextremal problems. In addition, they are characterized with superior performance properties. Nevertheless, each of the selected heuristic algorithms has demonstrated its own specific search features that allow the detection and identification of both global and local extremes. In this paper, the investigated algorithms have been validated on a larger test functions environment with different types of extremes. The particular attention was given to analyse their individual methods when solving the data-clustering problem. The main conclusion is that each of these methods can find the extremes by satisfying any desired precision and have acceptable performance, when applied to the variety of practical problems.

Pages: 86 to 99

Copyright: Copyright (c) to authors, 2017. Used with permission.

Publication date: June 30, 2017

Published in: journal

ISSN: 1942-261x