文献类型：学位论文

中文题名：基于一致性滤波算法的传感器网络估计问题研究

英文题名：Survey of Wireless Sensor Networks Estimation Problem Based on Consensus-based Kalman Filtering Algorithm

作者：王帅[1];

机构：[1]华东理工大学;

导师：侍洪波;华东理工大学

授予学位：硕士

语种：中文

中文关键词：分布式滤波;一致性算法;传感器网络;牵制控制;UKF

外文关键词：Distributed filtering;Consensus algorithm;Sensor network;Contain control;UKF

摘要：无线传感器网络(Wireless Sensor Network)是当前一个前沿的热点研究领域,有着广泛的应用前景。由于单传感器自身的能量、存储和处理能力的局限性,带来了稳定性差、可靠性较低、检测精度不高的挑战性问题,信息融合技术成为解决上述问题的有效手段,也引起了学者的极大关注。信息融合技术是一个多学科高度集成的热点研究领域,本文针对无线传感器网络自身特点,将无线传感器网络和信息融合算法相结合,对一致性滤波融合算法进行了研究。 论文提出了一种基于一致性的分布式滤波算法,针对实际应用中存在的网络丢包问题,重点研究了有丢包时的分布式滤波算法,通过理论分析给出了估计误差系统收敛的充分条件。通过仿真实验将本文算法与前人滤波算法在理想状况和有丢包状况下进行比较研究,表明本算法具有较优的滤波效果；探讨了丢包率对算法的影响,提出了三种改进方案,仿真表明每种方案都能较好地改善丢包问题；在保持一定精度条件和降低网络能耗的前提下,将该算法引入牵制网络结构中,讨论了牵制度大节点和牵制随机节点两种牵制策略,并通过仿真实验比较相应策略的跟踪精度及网络同步化性能；针对非线性系统,提出了一种基于一致性的UKF算法,仿真结果表明与传统EKF和UKF相比,一致性UKF具有良好的效果。
Wireless sensor networks /(WSN/) is currently a top research field with a wide range of applications. Since the limitations of single sensor's energy, storage and processing bring challenging issues including low stability, poor reliability and precision, information fusion technology can be an effective approach to address these problems and has caused great concern of scholars. Information fusion technology is a highly integrated multi-disciplinary research field and in this paper, for the characteristics of wireless sensor networks, we study the Kalman filter fusion algorithm based on consensus by combining wireless sensor networks with information fusion algorithm. In the first place, this paper introduces a consensus-based distributed filtering algorithm. Aiming to the problem of network packet-dropping in the practical application, we focus on the distributed filtering algorithm with packet-dropping. By the theoretical analysis, we give a sufficient condition for the convergence of the estimation error system. We compare our algorithm with a classical filtering one by some simulations in ideal and packet-dropping cases, respectively. The results show that our algorithm does better filtering in the packet-dropping case. In response to this problem, three schemes of improvement are presented and simulation results show the effectiveness. Secondly, in order to reduce the network energy consumption under the circumstances of maintaining certain precision, the proposed algorithm is introduced into contain network structure. Two contain strategies are put forward, containing big degree nodes and random node. Their tracking precision and network synchronization are compared through simulation. Finally, a consensus-based UKF algorithm applying in nonlinear system is introduced. The result of comparing traditional EKF and UKF with our algorithm indicates the latter has good effect.