以模糊限制之代理人協商應用於PC設計

Application of Fuzzy Constraint-Based Agent Negotiation to Collaborative Design - A Case Study for PC Design

韓延康(Yen-Kang Han)

指導教授 ： 賴國華 博士

元智大學/資訊學院/資訊工程學系/碩士(2004年)

摘要

在過去，產品的功能與外觀幾乎都由研發人員所主控，經過這些年，由於生活愈來愈e化，消費者更在乎產品的外觀與使用感覺。因此產品的開發，不單只是研發部門的工作，而是研發、工業設計與行銷等部門之團隊合作。而一符合消費者需求的產品，需要由各個不同領域的專家，提出各自的看法，共同設計完成。但在此合作設計過程中，經常因為專業上之看法不同，而產生設計概念互相衝突，而如何能產生一有效機制，使這些衝突能夠消弭。無疑的，協商是一最有用且最實際之方式。在現實生活中，各行各業最常用的協商方式就是開會。但除了開會之外，是否有更有效的方法來代替開會，因此，就有多不同的協商機制，應用於合作設計中。
在本論文中，提出一分散式模糊限制為基礎的代理人協商架構，應用於合作設計中。並以PC設計為一案例，期望以代理人來模塑在PC開發過程中之所有參與者，以模糊限制來呈現參與開發者對於產品之不精準性及不確定性，進而將此產品設計完成。另外，更期望能藉由此架構，來改善PC設計過程，以提高公司的研發能力及產品競爭力。
最後，再進一步的分析本論文所提出的方法與其他利用不同協商機制應用於合作設計中之方法，針對設計出產品之成本進行評估，並比較其利潤。在此評估過程中，本論文採用Bahler所提出的mixed quantitative/qualitative方法及目前業界所採用的設計方法在相同產品上進行比較。如同所預期的，本論文所提出的方法相較於其他方法可得到較高之利潤。並且，本論文所提出之方法，從低階產品、中階產品，一直到高階產品，其利潤分別高於使用mixed quantitative/qualitative method 3.9 %、5.85 %及6.47 %，高於使用目前業界方法2.47 %、5.83 %及7.72 %。

關鍵字

合作設計 ； 分散式模糊限制 ； 代理人 ； 代理人協商

並列摘要

A design often requires collaborative working among multidisciplinary engineers. However, the conflicts are mostly arise among sharply diverse viewpoints in the collaborative design process. In such contexts, the key challenge raised by the collaborative design of complex artifacts is how to find an acceptable agreement that maximize the overall preference values of all participants and minimize the disparity among engineers and customers under incomplete and imprecise domain knowledge. Thus, collaborative design as agent negotiation for problem solving, in help of mediating the conflicts, and find ‘win-win’ agreements to meet the whole requirements of all parties.
In this thesis, we presents a general problem-solving framework for modeling fuzzy constraint-based agent negotiation, and its extensions to application to collaborative design for PC design in. Meanwhile, we exploit the fuzzy constraints to represent imprecision and uncertainty in PC design process.
Finally, the selection of proper two methods: mixed quantitative/qualitative method and ad hoc method. Mixed quantitative/qualitative method is proposed for support designers to resolve conflicts in the design process, which is a constraint-based approach. Ad hoc method by using current design technology of IT industry. We then compare the gain from a low-end product to a high-end product with these methods, and the simulating results show that our method offers higher gain than the mixed quantitative/qualitative method by 3.9%, when designing a low-end product, by 5.85 when designing a middle-end product, and by 6.47% when designing a high-end product. Furthermore, our method similarly offers higher gain than ad hoc method by 2.47%, when designing a middle-end product, by 5.83% when designing a middle-end product, and by 7.72% when designing a high-end product.

並列關鍵字

Agent Negotiation ； Collaborative Design ； Fuzzy Constraints ； Multi-agent Systems

參考文獻

[Antonsson 1997] Antonsson, E. K., and Otto. K. N., 1997. “Improving Engineering Design with Fuzzy Sets,” In Fuzzy Information Engineering: A Guided Tour of Applications, D. Dubois, H. Prade, and R. R. Yager, (Eds.), Johnson Wiley & Sons, ch.38, pp. 633-654.

[Cutkosky 1993] Cutkosky, M., Engelmore, R., Fikes, R., Gruber, W., Genesereth, T., Mark, J. Tenenbaum. and J. Weber, 1993. “PACT: An experiment in Integrating Concurrenting Engineering Systems,” IEEE Computer, Special Issue on Computer Support for Concurrent Engineering, Vol. 26, No. 1, pp. 28-37.

[Dubois 1994] Dubois, D., H. Fragier, and H. Prade, 1994. “Propagation and satisfaction of flexible constraints," in: R. Yager, L. Zadeh (Eds.), Fuzzy Sets, Neural Networks and Soft Computing, Van Nostrand Reinhold, New York, pp. 166-187.

[Faratin 2000] Faratin, P., C. Sierra, and N. R, Jennings, 2000. “Using Similarity Criteria to Make Trade-offs in Automated Negotiation,” Artificial Intelligence, Vol. 142, Issue 2, pp. 205-237.

[Foroughi 1995] Foroughi, Abbas, 1995. “A Survey of the Use of Computer Support for Negotiation,” Journal of Applied Business Research, pp. 121-134.