Engineering Sciences for Modeling and Simulation-Based Life-Cycle Engineering: An Engineering Design Framework Integrating Robust Optimization and Structured Simulation

该补助金为风险和不确定条件下的工程设计决策研究提供启动资金。 具体来说，研究将集中在设计评估和优化方法的设计决策过程中面对的设计偏好，环境的不确定性，风险和不完整的数据。 所提出的方法将集成的概念设计理论，鲁棒优化，有序优化，和控制理论的模拟实验的方法。 作为对研究工作的初步尝试，PI将检查设计性能和设计效用的关键问题，建立设计不确定性和风险的正式表征，并开发设计优化模块。 随着计算能力的迅速提高，基于仿真的方法已经成为可行的大规模复杂系统。 目前的一套解决方案的方法将扩大到包括一个混合程序，结合了传统的优化程序的好处与计算机模拟的灵活性。 将开发一种最优模拟时间分配算法，该算法可以在不确定性/风险条件下动态确定不同场景的最佳模拟长度，从而显着降低总计算成本。 通过对机械部件设计和大系统设计中的设计实例的详细研究，将进一步加强理论结果。 如果成功，该框架将为工程设计提供理论基础，这将允许进一步发展的设计评价，性能评估和设计优化。 评估和优化方法将为设计工程师提供关键工具，以捕获关键设计因素，分析不确定性的影响，并在其生命周期内优化设计。新的模拟方法将提供一个评估，信任寻求机制的设计评估，这也将有助于计算工具和方法，可用于一般的模拟问题。

This grant provides initiation funding for research on engineering design decision making under conditions of risk and uncertainty. Specifically, the research will focus on design assessment and optimization methodologies for design decision-making processes in the face of design preferences, environmental uncertainty, risk, and incomplete data. The proposed methods will integrate the notion of design theory, robust optimization, ordinal optimization, and a control-theoretic approach to simulation experiments. As an initial attempt toward the research endeavors, the PIs will examine critical issues in design performance and design utility, establish formal characterization of design uncertainty and risk, and develop design optimization modules. With rapidly increasing computational capability, simulation-based approaches have become feasible for large-scale complex systems. The current set of solution methodologies will be expanded to include a hybrid procedure that combines the benefits of traditional optimization routines with the flexibility of computer simulation. An optimal-simulation-budget-allocation algorithm will be developed, which can dynamically determine the best simulation lengths for different scenarios under conditions of uncertainty/risk and thus significantly reduce the total computational cost. The theoretic results will be further strengthened through a detailed study of design cases in both mechanical component design and large-scale system design. If successful, the framework will provide a theoretic foundation for engineering design, which will allow further development in design valuation, performance assessment, and design optimization. The assessment and optimization methodologies will provide design engineers with critical tools that capture key design factors, analyze the effects of uncertainty, and optimize the design over its life cycle. The new simulation methodologies will offer an evaluative, confidence-seeking mechanism for design evaluation, which will also contribute to the computational tools and methodologies available for general simulation problems.