A hyper-heuristic algorithm via proximal policy optimization for multi-objective truss problems

This paper proposes a hyper-heuristic evolutionary algorithm via proximal policy optimization, named HHEA-PPO, for solving multi-objective truss optimization problems. HHEA-PPO has a two-layer structure: a high-level strategy and low-level heuristics. The high-level strategy consists of proximal pol...

Full description

Saved in:
Bibliographic Details
Published in:Expert systems with applications Vol. 256; p. 124929
Main Authors: Yin, Shihong, Xiang, Zhengrong
Format: Journal Article
Language:English
Published: Elsevier Ltd 05.12.2024
Subjects:
Dynamic crowding distance
Hyper-heuristic algorithm
Multi-objective optimization
Proximal policy optimization
Truss optimization
Hyper-heuristic algorithm
Proximal policy optimization
Multi-objective optimization
Dynamic crowding distance
Truss optimization
ISSN:0957-4174
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper proposes a hyper-heuristic evolutionary algorithm via proximal policy optimization, named HHEA-PPO, for solving multi-objective truss optimization problems. HHEA-PPO has a two-layer structure: a high-level strategy and low-level heuristics. The high-level strategy consists of proximal policy optimization, while the low-level heuristics consist of ten predefined heuristic operators. During the iteration process, the high-level strategy selects the most promising low-level heuristic according to the state of the individuals and the population. To maintain the convergence and distribution of the external Pareto archive, a dynamic crowding distance mechanism is employed. HHEA-PPO is applied to eight multi-objective truss optimization problems and compared with thirteen state-of-the-art optimization algorithms in terms of success rate, average computation duration, and average fitness evaluations to evaluate its performance. The results show that HHEA-PPO has higher search efficiency and greater stability, demonstrating its ability to solve large-scale engineering design problems.
ISSN:0957-4174
DOI:10.1016/j.eswa.2024.124929