Design Optimization and Stochastic Analysis based on the Moving Least Squares Method
Design Optimization and Stochastic Analysis based on the Moving Least Squares Method
复制标题
基于移动最小二乘法的设计优化和随机分析
DOI：
--
复制
发表时间：
2005
期刊：
影响因子：
--
通讯作者：
T. Zeguer
中科院分区：
文献类型：
--
作者：
V. Toropov;U. Schramm;A. Sahai;Royston Jones;T. Zeguer
1. 摘要 许多工业设计优化和随机分析问题具有以下共同特征：（i）响应函数是通过昂贵的数值计算来评估的，（ii）函数值可能包含一定程度的数值噪声。在本文中，这些特征通过使用响应函数的高质量近似来解决，特别关注移动最小二乘法（MLSM）。通过将 MLSM 应用到驾驶员安全气囊充气过程的自动校准和鲁棒性评估来说明该技术。安全气囊动态响应的标定被表述为优化问题。目的是最小化实验测试和数值模拟结果之间的差异。一旦完成校准，就会对其鲁棒性进行评估，该评估使用与优化过程中相同的 MLSM 技术近似值。 2.
1. Abstract Many industrial design optimization and stochastic analysis problems have the following common features: (i) the response functions are evaluated as a result of expensive numerical computations, and (ii) function values may contain some level of numerical noise. In this paper these features are addressed by the use of high quality approximations of the response functions with a particular focus on the Moving Least Squares Method (MLSM). The technique is illustrated by the application of MLSM to the automatic calibration and robustness assessment of the driver’s airbag inflation process. Calibration of the dynamic response of the airbag is formulated as an optimization problem. The objective is to minimize the difference between the experimental test and numerical simulation results. Once calibration has been achieved, an assessment of its robustness is performed, which utilizes the same approximation by MLSM technology as in the optimization process. 2.