TRUST-TECH-enhanced differential evolution methodology for box-constrained nonlinear optimisation. (2017)

• Record Type: Journal Article
• Authors: Zhang, Xuexia; Chiang, Hsiao-Dong; Chen, Weirong
• Is Part Of: International journal of bio-inspired computation. Volume 10:Number 1(2017)
• Page Start: 1
• Page End: 11
• Publication Date: 2017
• Subjects: differential evolution -- DE -- TRUST-TECH -- global optimisation -- hybrid methods; Biologically-inspired computing -- Periodicals; Computational biology -- Periodicals; 572.0285
• Languages: English
• ISSNs: 1758-0366

Abstract

The differential evolution algorithm and its variants have been developed to solve box-constrained optimisation problems with encouraging results. However, differential evolution still suffers from its poor ability to zoom in on promising regions to find a high-quality or global optimal solution. The transformation under stability-retraining equilibrium characterisation (TRUST-TECH) methodology is a systematical and deterministic method to find a set of multiple local optimal solutions. This paper presents a TRUST-TECH-enhanced differential evolution methodology (TT-DEM) to improve the performance of differential evolution method. In the TT-DEM framework, a differential evolution method is carried out to identify promising regions containing a set of high-quality solution or even the global optimal solution, while TRUST-TECH exploits the identified promising regions to compute those high-quality optimal solutions. Following this framework, the original differential evolution (DE) and three adaptive DEs are enhanced by TRUST-TECH. Numerical studies are conducted on several benchmark functions with promising results.