SGER: Multi-Agent Diagnostics

SGER：多代理诊断

• 批准号: 9978923
• 负责人: Abhijit Deshmukh
• 金额: $ 10.84万
• 依托单位: University of Massachusetts Amherst
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1999
• 资助国家: 美国
• 起止时间: 1999-06-15 至 2001-11-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9978923&HistoricalAwards=false
• 关键词: SGER; Multi; Agent; Diagnostics

This grant provides funding for tile development of diagnostic tools for multi-agent systems. Specifically, this research will develop methods to evaluate rationality, convergence and op-timality of the social preference or group objective given the preferences, local goals and interaction mechanisms of individual agents. The developed tools will identify conditions under which the rationality, convergence and desired solution quality requirements are not satisfied by the multi-agent system. The methods will be tested in the context of engineering design decision making. The tools will be made available to the research community, via the WWW, to validate existing or planned agent-based applications. If successful, the results of this research will improve the performance and reliability of agent-based solutions for the control arid coordination of complex systems. The primary goal of this research is to develop methods for answering fundamental questions about the validity and quality of decisions made by distributed decision-makers. Identifying conditions under which multi-agent systems produce acceptable results will help to determine appropriate application areas for agent-based solutions.

这笔赠款为多代理系统诊断工具的开发提供资金。具体来说，本研究将开发方法来评估社会偏好或群体目标的合理性、收敛性和最优性，考虑到个体代理的偏好、局部目标和交互机制。开发的工具将识别多智能体系统无法满足合理性、收敛性和所需解决方案质量要求的条件。这些方法将在工程设计决策的背景下进行测试。这些工具将通过 WWW 提供给研究界，以验证现有或计划的基于代理的应用程序。 如果成功，这项研究的结果将提高基于代理的复杂系统控制和协调解决方案的性能和可靠性。这项研究的主要目标是开发方法来回答有关分布式决策者决策的有效性和质量的基本问题。确定多代理系统产生可接受结果的条件将有助于确定基于代理的解决方案的适当应用领域。