Multi-core sine cosine optimization: Methods and inclusive analysis. (February 2021)

• Record Type: Journal Article
• Authors: Zhou, Wei; Wang, Pengjun; Heidari, Ali Asghar; Wang, Mingjing; Zhao, Xuehua; Chen, Huiling
• Is Part Of: Expert systems with applications. Volume 164(2021)
• Publication Date: 2021-02
• Subjects: Sine cosine algorithm -- Salp swarm algorithm -- Grey wolf optimizer -- Levy flight strategy -- Global optimization; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2020.113974
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: A multi-strategy boosted Sine Cosine Algorithm named SGLSCA is proposed. Salp Swarm Algorithm and Grey Wolf Optimizer are introduced into the basic SCA. Levy flight strategy is also embedded into SCA. The superior performance of SGLSCA is confirmed over various advanced algorithms. SGLSCA is tested over benchmark functions and engineering optimization. Abstract: The Sine Cosine Algorithm (SCA) is a popular population-based optimization method, which has shown competitive results compared to other algorithms, and it has been utilized to tackle optimization cases in various domains. Despite popularity, the initial SCA suffers from minimalistic originality, mediocre performance, and shallow mathematical model. In fact, there is undoubtedly room for improvement in the structure of original SCA because it may face problems of lazy convergence and inertia to local optima. To relieve these drawbacks, this paper develops a new multi-core SCA named SGLSCA, which is combined with three strategies based on the patterns of Salp Swarm Algorithm (SSA), Grey Wolf Optimizer (GWO), and Levy flight (LF). Based on introducing the updating strategy of SSA and GWO, it is proposed to strengthen the exploration aptitude of the conventional SCA. Also, the SSA updating strategy aims to further update the population based on the best solution of SCA, while the GWO updating plan helps using the top three solutions of SCA. Also, the LF strategy is embedded to achieve the random individual walk during the history of the exploration and further augment the competence of SCA to avoid local optimal solutions. To substantiate the structure and results of the proposed multi-core SCA, which is entitled SGLSCA, it is compared against nine state-of-art algorithms, six improved SCA variants, and nine successful advanced algorithms on 34 benchmark functions selected from 23 benchmark functions and 30 IEEE CEC 2014 benchmark problems. Additionally, three practical, real-world engineering problems are considered. The final experimental results expose that the multi-core SGLSCA outperforms other optimizers including LSHADE-cnEpSin and LSHADE methods in terms of both convergence and optimality of solutions. A public repository will support this research at http://aliasgharheidari.com for future works and possible guidance.