Article ID Journal Published Year Pages File Type
5473005 Aerospace Science and Technology 2017 11 Pages PDF
Abstract
To make multi-objective reliability-based design optimization (MORBDO) more effective for complex structure with multi-failure modes and multi-physics coupling, multiple response surface method (MRSM)-based artificial neural network (ANN) (ANN-MRSM) and dynamic multi-objective particle swarm optimization (DMOPSO) algorithm are proposed based on MRSM and MOPSO algorithm. The mathematical model of ANN-MRSM is established by using artificial neural network to fit the multiple response surface function. The DMOPSO algorithm is proposed by designing dynamic inertia weight and dynamic learning factors. The proposed approach is verified by the MORBDO of turbine blisk deformation and stress with respect to fluid-thermal-structure interaction from probabilistic analysis. The optimization design results show that the proposed approach has the promising potentials to improve computational efficiency with acceptable computational precision for the MORBDO of turbine blisk deformation and stress. Moreover, Pareto front curve and a set of viable design values of turbine blisk are obtained for the high-reliability high-performance design of turbine blisk. The presented efforts provide an effective approach for MORBDO of complex structures, and enrich mechanical reliability design theory as well.
Related Topics
Physical Sciences and Engineering Engineering Aerospace Engineering
Authors
, , , ,