Artificial bee colony algorithm with a pure crossover operation for binary optimization. (February 2021)

• Record Type: Journal Article
• Authors: Xiang, Wan-li; Li, Yin-zhen; He, Rui-chun; An, Mei-qing
• Is Part Of: Computers & industrial engineering. Volume 152(2021)
• Publication Date: 2021-02
• Subjects: Artificial bee colony algorithm -- Pure crossover operation -- Frequency of perturbation -- Opposition-based learning -- Facility location; Engineering -- Data processing -- Periodicals; Industrial engineering -- Periodicals; 620.00285
• DOI: 10.1016/j.cie.2020.107011
• Languages: English
• ISSNs: 0360-8352

Abstract

Highlights: A new perturbation scheme is presented to control the frequency of perturbation of parameters. A pure crossover operation is introduced in the employed bees phase. Probability choice mechanism is removed from the onlooker bees phase. A multiple scout bees mechanism and opposition-based learning technique are employed in the scout bees phase. A comprehensive experimental study is carried out. Abstract: Artificial bee colony (ABC) algorithm is one of famous swarm intelligence approaches for continuous optimization. With the help of a solution transformation technique, it can evolve in continuous space but consequences belong to binary space. On the basis of binary ABC, a novel artificial bee colony algorithm (nABC for short) is first proposed for better solving the uncapacitated facility location problem (UFLP). In nABC, a pure crossover operation is proposed to improve information sharing quality and remove random perturbation of original search strategy in employed bees phase. Next, a new frequency of perturbation is presented for enhancing the scale of information sharing between different individuals. Then, a new search strategy without probability mechanism of basic ABC is introduced in the onlooker bees phase. To further balance the enhanced exploitation ability, the original strategy of randomly producing an individual is substituted with an opposition-based learning technique with multiple scout bees and the frequency of perturbation mechanism. To testify the effectiveness and the convergence performance of nABC, it is compared with basic ABC and other famous methods for solving fifteen UFLPs from OR-library. Experimental results demonstrate that the proposed nABC is superior to other state-of-the-art approaches in terms of solution accuracy, convergence speed and robustness.