Multi-section-Oriented Robust Algorithm for Unconstrained Global Optimization

Abstract

We develop a new hybrid method called multi-section-oriented artificial bee colony MSO-ABC algorithm, to solve complex unconstrained optimization problems. The method initially decomposes the search region into finite collection of subregions in which the best interval value of the objective function is found using ABC algorithm. ABC algorithm uses chaotic maps to express parameter adaption to improve the search performance and escape from local optima. Then, interval ranking rule defined with respect to the optimistic decision makers’ point of view is applied to find most promising subregion containing best objective value for further division and non-promising region(s) are discarded. The technique narrows the search region by recursive decomposition and enhances computational effort toward better solution. The process is repeated reducing the search space consecutively. The global/ close to global values are obtained in the form of an interval with negligible width. Validity of MSO-ABC is tested on benchmark functions followed by the non-parametric tests viz., Friedman and Wilcoxon rank test for assessing statistical significance of the proposed method against other well-established optimization methods.