A hybrid algorithm based on self-adaptive gravitational search algorithm and differential evolution. (15th December 2018)

• Record Type: Journal Article
• Authors: Zhao, Fuqing; Xue, Feilong; Zhang, Yi; Ma, Weimin; Zhang, Chuck; Song, Houbin
• Is Part Of: Expert systems with applications. Volume 113(2018)
• Page Start: 515
• Page End: 530
• Publication Date: 2018-12-15
• Subjects: Gravitational search algorithm -- Differential evolution -- Self-adaptive mechanism -- Crossover and mutation operation -- Exploration and exploitation; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2018.07.008
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: A hybrid algorithm based on self-adaptive mechanism is proposed. A parameter self-adaptive mechanism based on α and G 0 is presented. A disturbance operation based on Levy flight is introduced. Two mutation mechanisms and crossover operation are proposed in the algorithm. The convergence performance of the algorithm is analyzed with the Markov model. Abstract: The Gravitational Search Algorithm (GSA) has excellent performance in solving various optimization problems. However, it has been demonstrated that GSA tends to trap into local optima and are easy to lose diversity in the late evolution process. In this paper, a new hybrid algorithm based on self-adaptive Gravitational Search Algorithm (GSA) and Differential Evolution (DE) is proposed for solving single objective optimization, named SGSADE. Firstly, a self-adaptive mechanism based on GSA is proposed for improving the convergence speed and balancing exploration and exploitation. Secondly, the diversity of the population is maintained in the evolution process by using crossover and mutation operation from DE. Besides, to improve the performance of the algorithm, a new perturbation based on Levy flight theory is embedded to enhance exploitation capacity. The simulated results of SGSADE on 2017 CEC benchmark functions show that the SGSADE outperforms the state-of-the-art variant algorithms of the GSA.