Title: Adaptive surrogate modelling algorithm for meta-model-based design optimisation

Authors: M.N.P. Meibody; H. Naseh; F. Ommi

Addresses: Aerospace Research Institute, Ministry of Science, Research and Technology, Tehran, Iran ' Aerospace Research Institute, Ministry of Science, Research and Technology, Tehran, Iran ' Department of Mechanical Engineering, Tarbiat Modares University, Tehran, Iran

Abstract: In this paper, an adaptive meta-modelling algorithm is proposed for complex systems surrogate modelling. Progressive Latin hypercube sampling (PLHS) has been developed as the design of experiments (DOE) method for meta-modelling. In this DOE, the number of samples increases in an iterative process until the meta-modelling accuracy converges. To evaluate the effects of design parameters on the system response, sensitivity analysis has been performed. Particle swarm optimisation (PSO) algorithm is applied as the optimiser. The proposed methodology reduces the computational costs of the design optimisation process. The PLHS-based surrogate modelling is applied to the design of a space thruster nozzle as a case study. In this case, propulsion efficiency and mass (key factors of space propulsion systems) are considered as objective functions.

Keywords: surrogate modelling; progressive Latin hypercube sampling; PLHS; meta-model-based design optimisation; MBDO; Kriging; space nozzle.

DOI: 10.1504/IJISE.2021.119721

International Journal of Industrial and Systems Engineering, 2021 Vol.39 No.3, pp.394 - 410

Received: 01 Oct 2019
Accepted: 15 Nov 2019
Published online: 16 Dec 2021 *

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