Validation of an automated kriging-based methodology to calibrate PSO parameters: application to parametric optimization of truss structures

Abstract

For years, the Particle Swarm Optimization (PSO) algorithm has been widely studied and many improved versions have been developed: from the swarm's topologies to the addition of new parameters, including machine learning approaches. However, the tuning of the fundamental PSO parameters has been less studied, but may lead to significant improvements on the convergence accuracy of PSO. This paper aims to develop an automated methodology to calibrate PSO parameters for a given optimization problem. The process is based on the kriging estimation of the best combination of PSO parameters. In this way, the Automated Tuning parameter Calibration (ATpC) methodology gives the optimal PSO setup for each considered problem in order to lead to a better convergence accuracy. The proposed ATpC methodology is applied to parametric optimization of truss structures. ATpC methodology performance is assessed by comparison of two different PSO setups usually used in the literature. The numerical results show that the ATpC methodology allows to significantly improve the convergence accuracy of PSO.