STRUCTURAL OPTIMIZATION BASED ON RESPONSE SURFACE METHODOLOGY
ZHANG Lixin SUI Yunkang (Numerical Simulation Center for Engineering, Beijing University of Technology, Beijing 100022, China) GAO Xueshi (Mechanical and Electronic Engineering College, University of Petroleum, Dongying 257062, China) FU Zhiyuan (Hengda Company, Sichuan Petroleum Administration, Zigong 643000, China)
Using response surface methodology (RSM), the responses of a system can be expressed in the explicit form of the design variables. For fitting response surfaces, the central symmetric design and the quasi-simplex design are developed according to central composite design and simplex design, respectively, in an attempt to cut down the computational cost involved in repetitive structural analyses with the accuracy being guaranteed. For a common optimization problem, the RSM can formulate it into an optimal model linearly or nonlinearly constrained. The optimization toolbox of Matlab can be employed to find the optima. The crashworthiness design shows that the application of the proposed experimental design methods and optimal model can achieve good accuracy and efficiency.