Enhanced Jaya algorithm: A simple but efficient optimization method for constrained engineering design problems

Yiying Zhang, Aining Chi, Seyedali Mirjalili

Research output: Contribution to journalArticlepeer-review

15 Citations (Scopus)

Abstract

Jaya algorithm (JAYA) is a new metaheuristic algorithm, which has a very simple structure and only requires population size and terminal condition for optimization. Given the two features, JAYA has been widely used to solve various types of optimization problems. However, JAYA may easily get trapped in local optima for solving complex optimization problems due to its single learning strategy with little population information. To improve the global search ability of JAYA, this work proposes an enhanced Jaya algorithm (EJAYA) for global optimization. In EJAYA, the local exploitation is based on defined upper and lower local attractors and global exploration is guided by historical population. Like JAYA, EJAYA does notneed any effort for fine tuning initial parameters. To check the performance of the proposed EJAYA, EJAYA is first used to solve 45 test functions extracted from the well-known CEC 2014 and CEC 2015 test suites. Then EJAYA is employed to solve seven challenging real-world engineering design optimization problems. Experimental results support the strong ability of EJAYA to escape from the local optimum for solving complex optimization problems and the effectively of the introduced improved strategies to JAYA. Note that, the source codes of the proposed EJAYA are publicly available at https://ww2.mathworks.cn/matlabcentral/fileexchange/88877-enhanced-jaya-algorithm-for-global-optimization.

Original languageEnglish
Article number107555
JournalKnowledge-Based Systems
Volume233
DOIs
Publication statusPublished - 5 Dec 2021

Keywords

  • Evolutionary computing
  • Global optimization
  • Jaya algorithm
  • Metaheuristics
  • Swarm intelligence

Fingerprint

Dive into the research topics of 'Enhanced Jaya algorithm: A simple but efficient optimization method for constrained engineering design problems'. Together they form a unique fingerprint.

Cite this