Method of Distributed Model Management and Combination Based on Multi-Agent

摘要

To satisfy requirements of making scientific decision quickly and correctly in distributed network environment, a Multi-agent Model System (MAMS)was proposed which could implement distributed model management and combination. The structures of MAMS constituents-management agent and model agent were designed, and their functions and actions were analysed. The paper utilizes reverse reasoning technology to combine model agents dynamically and automatically and a process was then provided which applied several model agents to solve a complex problem. Finally, taking a decision making in product development as an example, the working process of MAMS was discussed. Compared with traditional model systems, MAMS endowed with functions of model remote calling & operation, automatic combining, which provided a new approach to build model system for decision supporting system in resolving complex problems.