项目名称: 多自主体系统分布式优化滤波问题研究
项目编号: No.61203158
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 自动化学科
项目作者: 杨文
作者单位: 华东理工大学
项目金额: 24万元
中文摘要: 如何设计有效的分布式滤波算法使大规模的自主体(如传感器)之间通过协同合作对复杂/危险环境(如自然灾害、战役、沙漠、深海)中的目标进行状态估计、监测与搜救是一个重要而又困难的课题。本项目将深入研究网络拓扑结构、网络系统参数与分布式估计精度之间的关系,以最小化网络估计误差协方差或最大化估计速度为目标,基于优化、复杂网络等理论,设计优化的一致滤波算法。针对传感器网络能量有约束等情况,设计调度策略来调整部分传感器进行目标监测或调控网络拓扑结构,提出优化估计机制以减少网络的能量损耗。进一步,考虑到实际应用中存在的网络丢包给估计精度带来的影响,将研究两类丢包问题,应用矩阵论和概率论等工具,分析丢包率与估计精度之间的关系。通过搭建小型无线传感器网络平台,对已有的理论结果进行实验验证,为算法的实际应用奠定基础。
中文关键词: 一致性;复杂网络;多自主体;分布式滤波;优化问题
英文摘要: Designing efficient distributed filtering algorithm for cooperative/coordianted large-scale multi-agent system(wireless sensor network) is crucial to state estimation, surveillance and rescue problem in complex and dangerous circumstance, such as, natural disaster, battle field, desert and deep sea. To minimize network estimation error or maximize convergence speed, we design optimized consensus-based filtering algorithm using convex optimization and complex network theory, based on investigating the relationship between network topology, system parameters and network estimation accuracy.For a sensor network with limited energy, we propose optimized estimation strategies by scheduling different subset of sensors to observe or regulating network topology. Furthermore, considering the influence of packet-dropping on estimation accuracy, we study two classes of packet-dropping problems using matrix analysis and probability theory. We also verify all the theoretical results by building wireless sensor network testbed for shedding light on the engineering applications.
英文关键词: consensus problem;complex networks;multi-agent systems;distributed filtering;convex optimization