GOALI: Robust Product Design Selection under Uncertainty and for Competitive Advantage

本研究的目的是（1）开发一个综合的框架和方法，以帮助产品设计的稳健选择，无论是单一产品还是产品线，考虑到测量客户偏好的不确定性，由于随着时间的推移不断变化的偏好的不确定性，竞争反应的不确定性，以及不可控的工程设计参数的不确定性，并在此框架基础上（2）构建并验证了一个原型决策支持系统。在所提出的方法中，产品/产品线设计选择被视为一个选择与两个主要阶段：设计方案生成和设计方案评价。 在第一阶段（设计模块），通用信息， 沿着工程设计公式中的不确定性，被送入多目标鲁棒优化，以生成Pareto鲁棒最优设计方案集。在第二阶段（市场模块），通过考虑客户的偏好和选择，为集合中的每个备选方案生成需求和市场份额估计。基于需求函数和通用性，成本函数计算每个生成的设计方案，并用于计算每个方案的净利润现值。将在两个模块之间进行递归过程，以利用模拟方法说明两个模块中不可控参数的不确定性，并获得一套强有力的建议。如果成功的话，这项研究的成果将推动工程设计、营销科学和管理等不同领域的技术发展。本次调查的结果将在工程设计自动化和营销科学界广泛传播， 融入马里兰州大学的课程，并将技术和工艺转移到工业中。拟议的研究还将为学生提供机会，学习涉及产品设计选择的商业和工程实践，以及如何通过基础研究提高这些实践的有效性。

The objective of this research is to (1) develop an integrated framework and methodology to aid robust selection of product designs, either a single product or a product line, taking into account the uncertainties in measuring customer preferences, uncertainties due to changing preferences over time, uncertainties in competitive reactions, and uncertainty in uncontrollable engineering design parameters, and based on this framework (2) construct and validate a prototype decision support system. In the proposed approach, the product/product-line design selection is viewed as a selection with two main stages: design alternative generation and design alternative evaluation. In the first stage (design module), commonality information, along with uncertainty in the engineering design formulation, is fed into a multi-objective robust optimization to generate a Pareto robust optimal set of design alternatives. In the second stage (market module), demand and market share estimates are generated for each alternative in the set by accounting for customer preferences and choices. Based on the demand function and commonality, cost functions are calculated for each generated design alternative and are used to compute the net present value of profit for each alternative. A recursive process between the two modules will be undertaken to account for the uncertainty in uncontrollable parameters in both modules using simulation methods and to obtain a robust set of recommendations. If successful, the outcome of this research will advance the state of the art across different domains in engineering design, marketing science and management. The results of this investigation will be broadly disseminated in the engineering design automation and marketing science communities, the knowledge integrated into courses at the University of Maryland, and the technology and techniques transferred to industry. The proposed research will also provide opportunities for students to learn business and engineering practices involved in product design selection and ways to enhance the effectiveness of such practices through basic research.