A Comparison of Information Passing Strategies in System Level Modeling

• Main Authors: Ciucci, Francesco (Author), Lewis, Kemper (Author), Honda, Tomonori (Contributor), Yang, Maria C. (Author)
• Other Authors: Massachusetts Institute of Technology. Department of Mechanical Engineering (Contributor), Massachusetts Institute of Technology. Engineering Systems Division (Contributor), Massachusetts Institute of Technology. Institute for Data, Systems, and Society (Contributor), Yang, Maria (Contributor)
• Format: Article
• Language: English
• Published: ASME International, 2017-05-19T18:42:58Z.
• Subjects: Article
• Online Access: Get fulltext

Frameworks for modeling the communication and coordination of subsystem stakeholders are valuable for the synthesis of large engineering systems. However, these frameworks can be resource intensive and challenging to implement. This paper compares three frameworks, Multidisciplinary Design Optimization (MDO), traditional Game Theory, and a Modified Game Theoretic approach on the form and flow of information passed between subsystems. This paper considers the impact of "complete" information sharing by determining the effect of merging subsystems. Comparisons are made of convergence time and robustness in a case study of the design of a satellite. Results comparing MDO in two- and three-player scenarios indicate that, when the information passed between subsystems is sufficiently linear, the two scenarios converge in statistically indifferent number of iterations, but additional "complete" information does reduce variability in the number of iterations. The Modified Game Theoretic approach converges to a smaller region of the Pareto set compared to MDO, but does so without a system facilitator. Finally, a traditional Game Theoretic approach converges to a limit cycle rather than a fixed point for the given initial design. There may also be a region of attraction for convergence for a traditional Game Theoretic approach.
National Science Foundation (U.S.) (Award DMI-0547629)