基于D-S证据理论的无线传感器网络数据融合

发布时间：2018-03-04 14:09

本文选题：无线传感器网络　切入点：证据理论　出处：《北京交通大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着无线传感器网络技术的飞速发展和广泛应用,传感器网络的感知数据大量增加,如何实现传感网中的数据融合,以降低数据传输通信量、节约传感器节点耗能、延长WSNs工作寿命、提升数据处理精度成为目前无线传感器网络领域内的热点研究课题之一。论文首先对无线传感器网络决策融合问题以及证据理论基础知识进行了介绍,进而重点研究了 D-S证据理论在无线传感器网络多源数据融合过程中的若干关键问题。论文的研究工作得到了国家自然科学基金项目、北京市自然科学基金项目和高等学校博士学科点专项科研基金的支持。论文主要工作如下:(1)针对WSNs中故障节点检测问题,提出了一种基于传感器节点之间"测量值距离"的信度构建策略。在故障节点检测问题上,传统的BBA构建方法大多采用的是基于通信或读数是否正常的"次数"来建立,存在对未知信息利用不充分的缺点。本文提出的信度构建策略采用传感器与邻居节点之间的读数距离作为被测目标与样本数据间的距离计算方法,很好地利用了证据理论能够处理"不知道"信息的优点,并将传感器周围环境因素考虑进去,可有效地排除传感器受环境影响而产生暂歇性故障的类型,使得融合结果更为准确。论文还对其进行了仿真验证分析,仿真结果表明,网络故障率低于35%的情况下,本文提出的检测算法在故障检测率、漏检率以及误检率三个性能指标上所表现出的效果较好。(2)论文提出了一种解决证据理论融合过程中高冲突性证据问题的算法,采取"单个证据源具有不同的可信度"的思想,提出了一种以证据之间的"冲突量"作为对冲突信息分配的依据对冲突信息进行重新分配,将一部分冲突信息分配给证据的焦元,其余的分配给未知项m((?))。在无线传感器网络环境下,相比于已有的冲突处理算法,本文提出的组合规则具有融合过程收敛速度快、数据传输量更低、计算复杂度小的特点。论文采用了公式推导以及实例仿真的方法对新组合规则的正确性及功能性进行了验证分析,结果表明,本文提出的组合规则符合D-S证据理论的定义及性质,且其在融合收敛速度以及融合效果上都优于传统组合算法,并且本文算法具有更低的计算复杂度。论文的最后对研究成果进行了总结,并对后续的研究方向进行了展望。
[Abstract]:With the rapid development and wide application of wireless sensor network (WSN) technology, the perceptual data of sensor network is increasing greatly. How to realize data fusion in sensor network in order to reduce the amount of data communication and save the energy consumption of sensor nodes. Extending the working life of WSNs and improving the precision of data processing have become one of the hot research topics in the field of wireless sensor networks. Firstly, the decision fusion problem of wireless sensor networks and the basic knowledge of evidence theory are introduced in this paper. Furthermore, some key problems of D-S evidence theory in the process of multi-source data fusion in wireless sensor networks are studied. The research work of this paper has been carried out by the National Natural Science Foundation of China. The main work of this paper is as follows: (1) aiming at the problem of fault node detection in WSNs, In this paper, a reliability construction strategy based on the distance between sensor nodes is proposed. In the problem of fault node detection, most of the traditional BBA construction methods are based on whether the communication or reading is normal or not. The reliability construction strategy proposed in this paper uses the reading distance between sensor and neighbor node as the distance calculation method between the measured target and the sample data. It makes good use of the advantage of evidence theory to deal with the "unknown" information and take into account the environmental factors around the sensor, which can effectively eliminate the type of transient fault caused by the sensor being affected by the environment. The simulation results show that when the network failure rate is less than 35%, the detection algorithm proposed in this paper has a fault detection rate. This paper proposes an algorithm to solve the problem of high conflict evidence in the process of evidence theory fusion, which adopts the idea that "single evidence source has different credibility". In this paper, a new method of redistributing the conflict information based on the "conflict quantity" between the evidence is proposed, which distributes part of the conflict information to the focal element of the evidence, and the rest to the unknown item. In the wireless sensor network environment, compared with the existing conflict processing algorithms, the proposed combination rules have the advantages of fast convergence speed and lower data transmission. The method of formula derivation and example simulation is used to verify and analyze the correctness and functionality of the new combination rule. The results show that, The combination rule proposed in this paper accords with the definition and property of D-S evidence theory, and it is superior to the traditional combination algorithm in convergence speed and fusion effect. And this algorithm has lower computational complexity. Finally, the research results are summarized, and the future research direction is prospected.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP212.9;TN929.5

【参考文献】