Expensive Multimodal Simulation Optimization via Surrogate Model-Driven Particle Swarm Optimizer

View at publisher Summarise with VIRAT
Authors: Caitong Yue, Zekang Li, Qi Wei, Mingyuan Yu, Mengmeng Li, Jing Liang
Cited by: 0
References: 46

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Many real-world optimization problems are multimodal with multiple globally optimal solutions, yet their evaluations are often computationally expensive—defining expensive multimodal optimization problems. Surrogate-assisted evolutionary algorithms reduce costs but often fail to balance global optima discovery with solution precision under limited budgets. To address these challenges, this article proposes surrogate-assisted particle swarm algorithm (SAPSO). Its three key innovations are: first, a ring-topology global-local search framework that simultaneously refines known solutions and explores new regions; second, an archive selection strategy to eliminate redundant searches and prioritize unexplored promising regions; and finally, an improved multimodal detection strategy that identifies distinct optima using existing data without additional expensive evaluations. To validate its practical utility, SAPSO is first applied to optimize the parameters of a permanent magnet synchronous motor, yielding three reliable solutions for different operational needs. Furthermore, the extensive experiment is carried out on the CEC2013 test set. In terms of the global optimal solution indicators, SAPSO outperformed MaMPSO in 10 functions, DSCPSO-MMO in 19 functions, and MHS-QPSO in 11 functions. For the valley ratio, SAPSO achieved better results than MaMPSO in 12 functions, DSCPSO-MMO in 18 functions, and MHS-QPSO in 17 functions. These results demonstrate SAPSO’s ability to consistently discover multiple high-quality solutions.

Fields of Study & Subjects

Metaheuristic Optimization Algorithms Research Advanced Multi-Objective Optimization Algorithms Simulation Techniques and Applications

Publication details

Journal / source
IEEE Transactions on Industrial Informatics
Volume
22
Issue
7
Pages
5807–5818
Publication date
30 March 2026
Language
EN