An Exploration of Risk, Information, and Model Validity in Engineering Design

为这个项目进行的研究活动集中在调查基于决策的设计（DBD）要在设计研究人员社区中获得普及必须解决的主要问题。DBD是一个有争议的猜想，它断言工程设计是一个决策过程，因此可以在决策理论的既定原则框架内。这一猜想是建立在工程设计实践直接类似于理性经济行为的前提之上的。然而，到目前为止，DBD还不是一个可行的规范理论，因为它的公理化基础还没有导致后续分析方法的发展。虽然研究人员可以对不确定性下的设计提出许多问题，但该项目将首先关注是否有可能，以及随后的实践，将设计备选空间映射到满足DBD公理的领域。为了回答这个问题，任务将包括构建捕获设计备选方案（设计参数和设计配置）的概率空间的调查。这个问题必须解决，以评估模型的有效性；正如研究小组所认为的那样，有效性不仅仅是简单地满足理性公理。从这一努力中获得的结果的预期影响将是在与工程模型和预期效用定理相关的基本概率空间的构造之间建立有效的逻辑联系。这样的结果将在有效的工程设计空间内提供不确定性元素的直接分析表征。研究人员还将探索与德克萨斯大学奥斯汀分校的一组教师的合作研究活动。此次合作将调查设计空间概率建模工作与德克萨斯大学正在进行的经验相似研究中产生的不确定性表征之间的可能联系。

The research activities to be conducted for this project focus on investigating principal questions that must be resolved if Decision Based Design (DBD) is to gain currency within the community of design researchers. DBD is a controversial conjecture that asserts that engineering design is a decision making process and thus can be framed within well-established tenets of decision theory. This conjecture is founded on the premise that the practice of engineering design is directly analogous to rational economic behavior. As yet, however, DBD is not a viable normative theory in that its axiomatic foundation has not lead to the development of subsequent analytical methods. While researchers can pose numerous questions regarding designing under uncertainty, this project will focus first on examining whether it is possible, and subsequently practical, to map spaces of design alternatives into a domain that satisfies the axioms of DBD. To answer the question, tasks will include the investigation of constructing probability spaces that capture design alternatives (both design parameters and design configurations). This question must be addressed to assess model validity; as the research team holds that validity is more than simply satisfying axioms of rationality. The expected impact of the results attained from this effort will be to establish a valid logical connection between constructs of a basic probability space associated with engineering models and the expected utility theorem. Such a result will offer direct analytical characterization of elements of uncertainty within a valid engineering design space. The researchers will also explore a collaborative research activity with a group of faculty at the University of Texas at Austin. This collaboration will investigate possible connections between this work in probability modeling of design spaces with the characterization of uncertainties arising in the empirical similitude research being conducted at the University of Texas.