NSF Young Investigator: Accounting for Uncertainty in Multidisciplinary Design and Manufacturing Optimization

9457179雷诺现有的多学科设计优化（MDO）技术一般不考虑系统设计和制造的差异。在本研究中，稳健的优化设计策略（即基于田口的）将嵌入到目前正在开发的MDO算法中，该算法利用人工神经网络和全局灵敏度进行系统逼近。作为这一努力的一部分，将对人工神经网络在建模设计和制造不确定性方面的应用进行调查。基于目前对设计空间映射人工神经网络的研究，认为人工神经网络可以为整个产品生命周期设计的方差建模提供一种理想的机制。为了使美国公司在先进的产品设计和制造方面保持竞争力，必须开发新技术，帮助设计人员在整个产品设计周期中做出决策。在这项研究中开发的设计工具和方法将有助于确保美国在先进产品设计和制造方面的竞争力。

9457179 Renaud Existing multidisciplinary design optimization (MDO) technologies do not generally account for system design and manufacturing variances. In this research robust design strategies for optimization (i.e., Taguchi based) will be embedded in MDO algorithms currently being developed which make use of both artificial neural networks and global sensitivities for system approximation. As part of this effort an investigation of the utility of artificial neural networks for modeling design and manufacturing uncertainties will be undertaken. Based on current investigations of artificial neural networks for design space mapping, it is thought that artificial neural networks may provide an ideal mechanism for modeling variance throughout a product's life cycle design. In order for the United States companies to remain competitive in advanced product design and manufacture, new technologies to aid designers in making decisions throughout the product design cycle must be developed. The design tools and methodologies to be developed in this research will help ensure the advancement of United States competitiveness in advanced product design and manufacture.