Optimization and artificial intelligence: An in-depth analysis of multi-objective optimization, sampling methods, and regression algorithms applied to structural design

Author(s)

Gomes, Guilherme Ferreira ; Bendine, Kouider ; Pereira, Joao Luiz Junho

• Document type: Journal article
• Source: MECHANICS BASED DESIGN OF STRUCTURES AND MACHINES; v. N/A, p. 28-pg., 2025-03-18.

Abstract

This study addresses the challenge of structural optimization in Formula SAE chassis, focusing on balancing lightweight design with structural integrity. By integrating parametric optimization with AIdriven metamodeling, the research compares four multi-objective optimization algorithms-Non-Sorted Genetic Algorithm II, Multi-objective Lichtenberg Algorithm, Multi-objective Sunflower Optimization, and Multi-objective Particle Swarm Optimization-aiming to minimize chassis mass and maximize stiffness. The results show that AI-driven metamodeling significantly reduces computational cost, cutting optimization time by over 99%, while maintaining accuracy comparable to direct Finite Element simulations. This work provides a framework for enhanced automotive and structural optimization. (AU)

• FAPESP's process: 22/10683-7 - Is my benchmark of datasets challenging enough?
• Grantee: João Luiz Junho Pereira
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• FAPESP's process: 23/10419-0 - Multi-objective optimal selection of benchmarking datasets for unbiased and efficient machine learning algorithm evaluation
• Grantee: João Luiz Junho Pereira
• Support Opportunities: Scholarships abroad - Research Internship - Post-doctor