| In recent years, research on Multi-Agent theory and technology has become one of the main focuses in the field of Distributed Artificial Intelligence. Its basic goal is to accomplish complex tasks by making each target, programming, knowledge and action respectively and cooperatively. How to guarantee agents interact with each other effectively and orderly is the key problems of MAS.In this thesis, the researcher proposes a multi-agent multi-issue negotiation model and applies this model into course scheduling system. In order to achieve this purpose, the researcher explores how to solve the problem of resources allocation by negotiation among multi-agents. The major work and findings of this research can be listed in the following paragraphs.(1) To construct a multi-agent multi-issue negotiation modelIn this part, the researcher constructs a multi-agent multi-issue negotiation model on the bases of extensive analysis of the multi-agent automatic negotiation system. Firstly, the researcher defines the formalization system, and discusses the negotiation protocol and the process of negotiation. Then the researcher puts forward the dynamic negotiation strategy on the bases of time, resources and preference strategies. Finally, the interactive algorithm in negotiation system is introduced.(2) To design and implement the intelligent course scheduling system based on the technology of multi-agent multi-issue automated negotiationBased on the model of multi-agent multi-issue automated negotiation, the researcher designs and implements the intelligent course scheduling system. This system implements the task of resources allocation by negotiation among multi-agents and it also concerns about teachers' expectations and improves teachers' satisfaction for the course schedule.(3) To put forward a multiple utility function for evaluating the intelligent course scheduling systemAiming at the attributes that different issues have different weights during multi-agent multi-issue negotiation, the researcher designs a multiple utility function for achieving the general target utility. This function combines the satisfaction evaluation function of teachers' time slots with the utility function of other discrete issues for the intelligent course scheduling system.In the last section of this thesis, according to the proposed frame of reference and negotiation technology, the researcher implements the intelligent course scheduling system by adopting JAVA language on the IBM Aglet platform. The correctness of the research design and the feasibility of the system are verified by evaluating the results of the intelligent course scheduling system. |
PDF Full Text Request