| INNOVATIVE DESIGN OPTIMIZATION STRATEGY FOR THE AUTOMOTIVE INDUSTRY |
| M. S. KIM1, D. O. KANG1, S. J. HEO2 |
1Institute of Design Optimization
2Kookmin University |
|
|
|
|
|
| ABSTRACT |
|
In order to effectively solve modern automotive design problems including the results of nonlinear FEA and multi-body dynamics, a progressive meta-model based design optimization is presented. To reduce the number of initial sample points, two sampling methods are introduced. Then, for efficient and stable construction of meta-models, three metamodel methods are newly introduced which are numerically based on the singular value decomposition technique. To design a practical system considering manufacturing tolerances and optimizing multiple performances, a robust design optimization, 6-sigma constraints and multi-objective strategies are implemented when solving the approximate optimization problem constructed from the meta-models. Until the convergence criteria are satisfied, the initially developed meta-models are progressively improved by adding only one point that minimizes the approximate Lagrangian in the consecutive optimization iterations. Finally, one validation sample and four automotive applications are solved to show the effectiveness of the proposed approach. |
| Key Words:
RDO (Robust Design Optimization), RBF (Radial Basis Function), RSM (Response Surface Method), MDO (Multi-Disciplinary Optimization), ALM (Augmented Lagrange Multiplier), DOE (Design of Experiments), SAO (Sequential Approximation Optimization), PMM (Progressive Meta-Model) |
|
|
|
Preview
Full text via DOI
Download Citation 
Print 
