贝叶斯网和MAS在生态工业园中的应用初探  

Application of Bayesian Network and MAS in Eco-industrial

作　　者：滕丽华[1] 

机构地区：[1]浙江万里学院生物与环境学院,浙江宁波315100

出　　处：《复杂系统与复杂性科学》2008年第1期81-86,共6页Complex Systems and Complexity Science

基　　金：浙江省青年教师资助项目(2005)

摘　　要：Agent在MAS中的行为在很多方面类似于生物在自然环境中的生态行为,因此,从生态系统角度,基于生态特征研究MAS的工业进化模型是很有意义的,对解决生态工业系统的优化控制有着引人注目的现实意义和发展前景。应用贝叶斯技术和贝叶斯网络为智能Agents和多Agent系统所提供的数学框架和处理工具,可有效地解决Agent研究领域所遇到的知识表示、学习和行为决策问题。生态工业系统中的每个企业可以建模为智能Agent。提出一种基于贝叶斯网的方法来对各企业Agent之间的协作关系进行建模。该方法能有效表达各企业Agent的行为策略之间的因果关系,进而可以对生态工业系统中各企业在不同情况下采取的行为策略进行预测和决策,使得整个生态工业系统的经济效益、社会效益和环境效益达到最优化。Agent in MAS in many ways is similar to those of the natural environment in the ecological behavior. Therefore, from the perspective of the ecosystem, based on the ecological characteristics of MAS industrial evolutionary model is very meaningful, eco-industrial system for solving optimal control of a striking relevance and prospects for future. Bayesian techniques and Bayesian network for smart Agents and Multi-Agent System provided by the mathematical framework and tools, which can effectively solve the Agent encountered by the research fields of knowledge expressing, learning and decision-making behavior. Every enterprise in eco-industrial system can be modeled as an Agent . In this paper, a method based on Bayesian network is proposed, the method can be effectively presented the casual relationship among enterprise Agents, which could be the ecological system of the industrial enterprises in different situations to predict the behavior and decision-making strategy, so that the whole system of eco-industrial economic, social and environmental benefits achieves optimization.

关 键 词：生态工业系统 AGENT系统 贝叶斯网 

分 类 号：TP399[自动化与计算机技术—计算机应用技术] N94[自动化与计算机技术—计算机科学与技术]