Chaos-assisted multi-population salp swarm algorithms: Framework and case studies. (15th April 2021)

• Record Type: Journal Article
• Authors: Liu, Yun; Shi, Yanqing; Chen, Hao; Heidari, Ali Asghar; Gui, Wenyong; Wang, Mingjing; Chen, Huiling; Li, Chengye
• Is Part Of: Expert systems with applications. Volume 168(2021)
• Publication Date: 2021-04-15
• Subjects: Salp swarm algorithm -- Chaos-assisted exploitation strategy -- Meta-heuristic -- Global optimization -- Swarm intelligence; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2020.114369
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: Multi-population mechanism is introduced into salp swarm algorithm. The proposed method is abbreviated as MCSSA. Chaos-assisted exploitation strategy is introduced into MCSSA. MCSSA is tested over 30 benchmark functions and 5 real-world problems. The greater effectiveness of MCSSA is confirmed over the advanced algorithms. Abstract: Salp swarm algorithm (SSA) is a recently presented algorithm, which is simple in structure and relatively mediocre in its performance. However, the original SSA still has features to be improved because it may face problems in convergence trends or easily being trapped into local optima for more advanced problems. To alleviate this limitation, we propose a new SSA-based method (MCSSA) that performs the chaotic exploitative trends and has a multi-population structure. The new structure can assist SSA in making a more stable tradeoff between global exploration and local exploitation capabilities. First, the exploitation trends and neighborhood searching commands of SSA are enriched using the chaos-assisted exploitation strategy. Next, we arrange a multi-population structure with three sub-strategies to augment the global exploration capabilities of the algorithm. To test the performance of this proposed MCSSA, a set of comprehensive algorithms is used, including 11 other original methods, conventional SSA, and 13 advanced techniques including SCA, SSA, GWO, MFO, WOA, BA, FPA, PSO, ALO, MVO, DE, ABC, CSSA, ESSA, CLSGMFO, LGCMFO, SaDE, jDE, EPSO, ALCPSO, CBA, RCBA, BWOA, CCMWOA, and GA-MPC based on 30 IEEE CEC2017 benchmark functions and 5 IEEE CEC2011 practical test problems. Also, the non-parametric statistics Wilcoxon signed-rank test and Friedman test are also used as an enabling tool to validate the performance of the proposed algorithm. From the result analysis, it can be concluded that the introduced strategy significantly improves the speed of the algorithm converging to the optimal value, and the improvement of the search ability also helps the algorithm to find a better solution than the basic SSA. As a conclusion, it can be said that MCSSA is reliable and efficient in solving complex optimization problems. An online website at https://aliasgharheidari.com supports this research for any guide or info.