Improved salp swarm algorithm combined with chaos

• Published in: Mathematics and computers in simulation Vol. 202; pp. 113 - 148
• Main Authors: Tawhid, Mohamed A., Ibrahim, Abdelmonem M.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.12.2022
• Subjects: Bio-inspired algorithm; Chaos theory; Metaheuristics; Nonlinear system of equations; Salp Swarm Algorithm; Metaheuristics; Chaos theory; Nonlinear system of equations; Bio-inspired algorithm; Salp Swarm Algorithm
• ISSN: 0378-4754; 1872-7166
• DOI: 10.1016/j.matcom.2022.05.029

A recently developed metaheuristic optimization algorithm, Salp Swarm Algorithm (SSA), has manifested its capability in solving various optimization problems and many real-life applications. SSA is based on salps’ swarming behaviour when finding their way and searching for food in the oceans. Nonetheless, like most metaheuristic algorithms, SSA experiences low convergence and stagnation in local optima and rate. There is a need to enhance SSA to speed its convergence and effectiveness to solve complex problems. In the present study, we will introduce chaos into SSA (CSSA) to increase its global search mobility for robust global optimization. Detailed studies are carried out on real-world nonlinear benchmark systems and CEC 2013 benchmark functions with chaotic map (Tent). Here, the algorithm utilizes a Tent map to tune the salp leaders’ attractive movement around food sources. The experimental results, considering both convergence and accuracy simultaneously, demonstrate the effectiveness of CSSA for 12 nonlinear systems and 28 unconstrained optimization problems CEC 2013. Two nonparametric statistical tests, the Friedman test and Wilcoxon Signed-Rank Test, are conducted to show the superiority of CSCA over other states of the art algorithms and our results’ significance. •This paper introduces chaos into Salp Swarm Algorithm.•Solve 28 unconstrained optimization problems, CEC and 12 nonlinear systems.•The effectiveness and efficiency of our algorithm are provided.•Experimental results prove superiority of our algorithm over the state-of-the-arts.