An improved bat algorithm with variable neighborhood search for global optimization

Abstract

Bat algorithm (BA) is newly proposed bio-inspired metaheuristic algorithm with the inspiration of the echolocation of bats in nature. Several experimental results have proven to the effectiveness and performance of BA. However, BA may fail to find the global optimal solution occasionally. In this paper, a kind of classical search technology, called variable neighborhood search (VNS), is incorporated into BA as a local search tool. An improved version of BA namely variable neighborhood bat algorithm (VNBA), is thus proposed. In VNBA, the classic BA as a global search tool searches the whole space globally, and this can significantly shrink the search space. Subsequently, VNS as a local search tool is implemented to find the final best solution within the small promising area. After that, the VNBA is benchmarked by sixteen standard benchmark functions. The experimental results imply that VNBA takes the absolute advantage over the basic BA.