Find-Fix-Finish-Exploit-Analyze (F3EA) meta-heuristic algorithm: An effective algorithm with new evolutionary operators for global optimization. (February 2019)

• Record Type: Journal Article
• Authors: Husseinzadeh Kashan, Ali; Tavakkoli-Moghaddam, Reza; Gen, Mitsuo
• Is Part Of: Computers & industrial engineering. Volume 128(2019)
• Page Start: 192
• Page End: 218
• Publication Date: 2019-02
• Subjects: Global optimization -- Evolutionary computation -- Selection, mutation and local search operators -- F3EA targeting process; Engineering -- Data processing -- Periodicals; Industrial engineering -- Periodicals; 620.00285
• DOI: 10.1016/j.cie.2018.12.033
• Languages: English
• ISSNs: 0360-8352

Abstract

Highlights: An evolutionary meta-heuristic algorithm is proposed and inspired by a targeting process. Algorithm exhibits a fast convergence behavior on engineering optimization problems. The algorithm affords from cleverly designed operators. The algorithm behaves constantly and performs more reliable than other algorithms. Abstract: A novel population-based evolutionary meta-heuristic algorithm is introduced, which imitates the Find-Fix-Finish-Exploit-Analyze (F3EA) targeting process. It considers the surface of the objective function as the battlefield and executes Find-Fix-Finish-Exploit-Analyze steps in an iterative manner. Following the radar detection rationale, a new evolutionary selection operator is introduced during the Find step. It is justified how to model the Fix step as a one-dimensional optimization problem to attain a local search operator. To produce a new solution by the Finish step, the target solution selected in the Find step is actioned artificially. This is an adaptive mutation stage, in which the position of the new potential solution is identified via modeling of projectile motion. The Exploit step takes over opportunities provided by mating the generated solution and its parent solution. Finally, the Analyze step, updates the population. Extensive experiments are conducted based on engineering optimization problems and a large set of benchmark functions for performance assessment, sensitivity analysis of the control parameters, and effectiveness analysis of different steps of the algorithm. Results of statistical tests signify that equipping the algorithm with new selection, mutation and local search operators makes it effective and efficient enough to exceed or match the best of rivals.