一种判定多智能体系统状态更新有序性的方法
发布时间：2018-07-22 16:50
【摘要】：多智能体模型的构建可分为若干环节,其中状态更新规则往往被设定为同步状态更新规则。然而,现实自然和社会系统中普遍存在的是异步状态更新规则。已有文献发现异步状态更新规则下和同步状态更新规则下的系统动力学过程有着显著区别,但目前尚实证或实验研究探索异步性的性质。为此,基于Pearson相关系数定义了状态更新序列间一致性度量指标,然后基于由随机异步状态更新规则产生的序列间一致性频率分布,构建了判定状态更新规则有序性的方法。最后,将所构建方法应用于一组二阶段任务更新实验数据,探索性分析了现实系统的异步状态更新规则性质。结果发现,大部分系统的状态更新规则具有随机有序性。研究结论可帮助理解社会系统动力学过程,同时对多智能体建模具有重要启示。
[Abstract]:The construction of multi-agent model can be divided into several steps, in which state update rules are often set as synchronous state update rules. However, asynchronous state update rules are common in real-world and social systems. It has been found that there are significant differences between the asynchronous state update rules and the synchronous state update rules in the system dynamics process, but there are still some empirical or experimental studies to explore the nature of asynchronism. Therefore, based on Pearson correlation coefficient, the consistency measure index between state update sequences is defined, and then based on the frequency distribution of consistency among sequences generated by random asynchronous state update rules, a method to determine the order of state update rules is proposed. Finally, the proposed method is applied to a set of two-stage task update experiment data, and the nature of asynchronous state update rules of the real system is analyzed. The results show that the state update rules of most systems are random and ordered. The conclusions can help to understand the process of social system dynamics and have important implications for multi-agent modeling.
【作者单位】： 上海对外经贸大学统计与信息学院;上海财经大学信息管理与工程学院;上海财经大学实验中心;驻上海航天局八0二所军代室;
【基金】：国家自然科学基金资助项目(71271126) 教育部人文社会科学研究规划项目(15YJCZH201) 2017年上海高校青年教师培养计划项目 上海市教育委员会科研创新项目(14YZ134)
【分类号】：TP18
[Abstract]:The construction of multi-agent model can be divided into several steps, in which state update rules are often set as synchronous state update rules. However, asynchronous state update rules are common in real-world and social systems. It has been found that there are significant differences between the asynchronous state update rules and the synchronous state update rules in the system dynamics process, but there are still some empirical or experimental studies to explore the nature of asynchronism. Therefore, based on Pearson correlation coefficient, the consistency measure index between state update sequences is defined, and then based on the frequency distribution of consistency among sequences generated by random asynchronous state update rules, a method to determine the order of state update rules is proposed. Finally, the proposed method is applied to a set of two-stage task update experiment data, and the nature of asynchronous state update rules of the real system is analyzed. The results show that the state update rules of most systems are random and ordered. The conclusions can help to understand the process of social system dynamics and have important implications for multi-agent modeling.
【作者单位】： 上海对外经贸大学统计与信息学院;上海财经大学信息管理与工程学院;上海财经大学实验中心;驻上海航天局八0二所军代室;
【基金】：国家自然科学基金资助项目(71271126) 教育部人文社会科学研究规划项目(15YJCZH201) 2017年上海高校青年教师培养计划项目 上海市教育委员会科研创新项目(14YZ134)
【分类号】：TP18
【相似文献】
相关期刊论文 前10条
1 李晓萌,杨煜普,许晓鸣;基于独立学习的多智能体协作决策[J];控制与决策;2002年01期
2 郭鲁;苏文明;;企业内组织的多智能体论述[J];科技广场;2008年02期
3 周，
本文编号：2138116
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2138116.html
