Collaborative Research: Visual Analytics for Creation of Value Functions in Complex Systems Design Under Uncertainty

价值驱动设计代表了一种设计大型复杂工程系统的新方法，如汽车、航空航天和能源行业中的系统。在价值驱动的设计中，创建了一个价值函数来捕捉利益相关者的偏好。价值函数包括对正在设计的系统非常重要的所有属性，例如重量、性能特征和制造问题等。然后，设计过程中的所有参与者都使用价值函数来帮助他们做出决策，以便所有参与者都使用相同的基础来进行设计决策。这与目前的系统工程过程截然不同，在当前的系统工程过程中，需求被用来迫使子系统和组件之间做出设计决策，从而导致次优设计。然而，对于复杂系统来说，创建价值函数是非常具有挑战性的，特别是考虑到设计变量、属性以及最终系统本身的价值中存在的不确定性。该奖项支持基础研究，为在不确定情况下设计复杂系统时创造价值函数提供必要的知识。这项研究的一个关键组成部分侧重于使用可视化工具轻松理解设计变量、属性和值函数之间的复杂关系。鉴于军事和商业应用对大规模系统的广泛依赖，这项研究的结果将有利于美国经济和国防。这项研究本质上是多学科的，不仅提供了一种在多学科环境中进行设计的方法，而且通过使用多学科来解决与复杂系统的价值功能创造相关的基本问题。这项研究的多学科性质将有助于扩大代表不足的群体在这项研究中的参与。价值驱动的设计方法可以克服与当前系统工程过程相关的许多限制。在目前的系统工程方法中，开发需求是为了获取利益相关者的偏好。然后，这些需求被分解并传递给较低级别的小组，这些小组然后设计子系统，并最终设计系统中的组件。有问题的是，这些需求过度限制了设计空间，在合并不确定性方面造成了困难，并且很少或根本没有提供利益相关者实际所需的指导。虽然价值驱动的设计有可能克服这些障碍，但知识差距仍然存在。本研究将解决不确定性条件下价值函数形成的知识鸿沟问题。研究小组将确定设计变量和属性中的不确定性对价值函数的影响。该小组将开发一种结构化的方法来形成价值函数，以确保根据利益相关者的偏好对备选解决方案进行适当的排序。本研究将探讨风险偏好(即风险寻求、风险中性和风险厌恶)的影响，以确定风险偏好与价值函数形成之间的关系。将开发可视化工具，以帮助理解属性和价值函数之间的关系，并为设计者提供使用价值函数的决策辅助。研究小组还将与行业合作伙伴合作，验证开发的方法和工具。

Value-driven design represents a new way of designing large scale complex engineered systems, such as those found in automobile, aerospace, and energy industries. In value-driven design, a value function is created that captures the preferences of the stakeholders. The value function comprises all the attributes that are important for the system being designed, such as weight, performance characteristics, and manufacturing concerns, amongst others. The value function is then used by all the participants in the design process to aid in their decision making, so that all participants use the same basis for their design decisions. This is extremely different from the present systems engineering process in which requirements are used to force design decisions among subsystems and components, leading to suboptimal designs. The creation of a value function, however, is very challenging for complex systems, particularly in light of the uncertainties that exist in the design variables, attributes, and ultimately, the value of the system itself. This award supports fundamental research to provide needed knowledge for the creation of value functions when designing complex systems under uncertainty. A key component of this research focuses on enabling the easy understanding of the complex relationships between design variables, attributes, and value functions using visualization tools. Results from this research will benefit the U.S. economy as well as the national defense, given the extensive reliance on large-scale systems by both military and commercial applications. This research is inherently multidisciplinary, not only providing a way to design in a multidisciplinary environment, but also through using multiple disciplines to address the fundamental issues related to value function creation for complex systems. The multidisciplinary nature of the research will lend itself to broadening participation of under-represented groups in this research.The value-driven design approach can overcome numerous limitations associated with present systems engineering processes. In the present systems engineering approach, requirements are developed in an attempt to capture the preferences of stakeholders. These requirements are then decomposed and passed to lower-level groups who then design subsystems and ultimately the components in the system. Problematically, these requirements over-restrict the design space, create difficulties in incorporating uncertainty, and provide little to no guidance in terms of what is actually desired by the stakeholders. While value-driven design has the potential to overcome these barriers, knowledge gaps exist. This research will address the knowledge gap related to formation of the value function under uncertainty. The research team will identify the impact of uncertainty in design variables and attributes on value functions. The team will develop a structured approach for forming a value function to ensure appropriate rank ordering of alternative solutions according to the preference of the stakeholder. The impact of risk preference (i.e., risk seeking, risk neutral, and risk averse) will be explored to determine relationships between risk preference and value-function formation. Visualization tools will be developed to aid in the understanding of relationships between attributes and the value function and to provide designers with decision aids for using value functions. The research team will also work with industrial partners to validate the methods and tools developed.