Title: Dual model surrogate-assist evolutionary algorithm for expensive multi-objective optimisation

Authors: Songyi Xiao; Wenjun Wang

Addresses: School of Automation, Guangdong University of Technology, Guangzhou, 510006, China ' School of Business Administration, Nanchang Institute of Technology, Nanchang, 330099, China

Abstract: Surrogate-assisted evolutionary algorithms (SAEAs) efficiently solve expensive optimisation problems by employing surrogate models to identify promising solutions. However, the surrogate model's weak performance in a limited sample prevents accurate prediction of solution fitness and the identification of promising solutions. To overcome these limitations, we propose a dual model-based surrogate-assisted evolutionary algorithm (DM-SAEA). The dual model consists of a Gaussian model used as the global model and a pairwise ranker employed as the local model. Meanwhile, an offspring selection strategy is proposed to select promising solutions based on the cooperation of dual models. Moreover, a dynamic fitness function is developed based on the Pareto rank and penalty boundary intersection to enhance the discriminatory quality of the solutions. The experimental results demonstrate that the proposed DA-SAEA outperforms current evolutionary algorithms in addressing various expensive multi-objective test problems.

Keywords: dual surrogate model; expensive multi-objective problem; pairwise ranker; offspring selection.

DOI: 10.1504/IJBIC.2024.139276

International Journal of Bio-Inspired Computation, 2024 Vol.23 No.4, pp.236 - 244

Received: 19 Dec 2023
Accepted: 08 Mar 2024
Published online: 28 Jun 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article