The Artificial Bee Colony (ABC) algorithm is a relatively new member of swarm intelligence based algorithms, which had shown better performance than other population based algorithms or the same performance; however this algorithm has deficiency in deal with some functions. This paper presents two appropriate local search heuristics as complementary of the ABC algorithm. In addition to present these local search heuristics, a parallel approach is proposed to select best values achieved by local searches. This approach improves robustness of ABC with local search, and consequently makes it more generally applicable than ABC or ABC with specific local search. The proposed algorithm is tested on nine benchmark functions which are hard for ABC. Numerical results show that these local searches obtain high-quality results in some problems, also according to problem properties, one local search may operate more successful than the others, and with parallel auto select method higher quality of solution can be achieved compare to use of specific local search.
展开▼
