Stochastic Strategies for Distributed Decisions in Concurrent Engineering

本研究：(1)将设计团队其他成员提出的“设计概念”的变化视为“概念噪音”；(2)展示了如何将这些噪声纳入“概念上稳健”的决策；(3)描述了一种使用相同分析向其他团队成员提供偏好信息的方法；(4)提供了确定是否发布概念上稳健的设计或等待进一步的设计确定性的程序。田口的方法被用作起点，因为它广为人知。本课题采用遗传算法对并发分布式决策过程进行了推广。并行工程过程涉及多功能团队，他们同时对产品生产系统的许多部分和产品生命周期的各个方面做出决策。这项研究认为，这种并发的分布式决策必须基于关于可能性集合的通信，而不是单一的解决方案。通过扩展田口的参数设计概念，它基于这种通信开发了一个健壮的分布式决策过程。也就是说，它显示了设计团队的成员如何根据来自团队其他成员的不完整信息做出适当的决策。

This research: (1) treats variations in the "design concepts" posed by other members of the design team as "conceptual noise"; (2) shows how to incorporate such noise into "conceptually robust" decisions; (3) describes a method for using the same analysis to provide preference information back to the other team members; and (4) provides a procedure for determining whether to release the conceptually robust design or to wait for further design certainty. Taguchi's approach is used as a starting point because it is widely known. This project generalizes concurrent distributed decision making procedure by employing genetic algorithms. Concurrent engineering processes involve multi-functional teams, which make decisions about many parts of the product\production system and aspects of the product life-cycle simultaneously. This research argues that such concurrent distributed decisions must be based on communications about sets of possibilities rather than single solutions. By extending Taguchi's parameter design concepts, it develops a robust and distributed decision-making procedure based on such communications. That is, it shows how a member of a design team can make appropriate decisions based on incomplete information from the other members of the team.