【作者】 张云洲；

【导师】 薛定宇； 吴成东；

【作者基本信息】 东北大学 ， 模式识别与智能系统， 2009， 博士

【摘要】 随着社会与经济的发展,大型建筑的数量迅速增多,其结构更加复杂,对火灾、爆炸等突发灾难的救援提出了更高的要求。无线传感器网络(Wireless Sensor Networks, WSN)具有不依赖基础设施、分布式数据结构、更高的可靠性等优点,为大型建筑灾难救援提供了崭新的研究手段。本文面向大型建筑的火灾等突发灾难,在传统总线型火灾报警系统的基础上,引入无线传感器网络,探索建立具有精确定位能力、能够快速准确传递信息、具有更强生命力和更高效率的大型建筑突发灾难救援系统,论文对其中的若干关键技术进行了理论探索和应用研究。首先,研究开发了基于现场总线的大型建筑智能火灾报警系统,采用高性能微处理器和多种通信协议相结合的模块化设计方案,使系统具备了良好的扩展能力,可以适应不同规模建筑物的需求。同时开发了面向灾难救援的便携式信息终端,能够使救援人员及时获取现场信息,并实现与指挥中心的多媒体综合信息交互,从而与指挥员建立通畅的联系,有助于提高救援效率。其次,研究基于无线传感器网络的着火点定位和多目标定位方法。在高斯空间多峰火灾模型的基础上,对基于能量的多目标定位算法进行了改进,提出了一种新颖的EM-MR算法,接着将禁忌搜索算法用于目标位置参数估计,成功实现了多着火点位置参数的准确搜索,在提高定位精度的同时有效降低了计算复杂度；建立了多声源定位的模型,分别将所提出的改进MR算法和禁忌算法应用于多声源定位,通过仿真实验证明了所提算法的有效性。再次,对含障碍物区域内的无线传感器网络确定性部署问题进行了研究。建立了传感器节点的探测模型和网络部署质量评价方式；基于考虑误警率的概率探测模型,提出了一种新的部署方法,先采用分水岭算法选取新增节点的部署区域,再以Delaunay剖分算法产生新增节点的候选部署位置。结合合理的节点部署位置评价机制,实现了传感器节点的有序高效部署。与随机部署、MAX_MIN_COV和MAX AVG COV等算法相比,所实现的部署策略在节点数有限的情况下,能够实现更高的覆盖探测概率和更好的覆盖一致性。由于建筑灾难救援系统对信息的快速、准确传递有严格的要求,本文采用簇内的数据融合树结构,以顶点覆盖技术为基础,形成最小转发最短路径数据融合树结构,并根据树形结构动态分配融合节点的融合时间,减少数据的传输时延,避免数据传输过程中产生的冲突。应用图理论及聚类技术,提出了一种以分簇协议及顶点覆盖技术为基础的簇内树状模糊C-均值分布式异类数据融合算法,并通过仿真实验证明了算法的效率。最后,考虑现有节点均存在不同程度的技术保密、无法完全开展路由协议和特定信号探测研究的实际情况,基于ZigBee协议开发了一种新型的无线传感器网络节点。在分析ZigBee的协议体系结构及协议各层作用的基础上,采用微处理器和ZigBee器件设计了节点硬件,将基于Micaz硬件平台的TinyOS系统移植到自主研发的节点硬件平台上；利用TinyOS系统编程实现了节点各模块功能,并开发了由多个网络节点组成星型网络的应用程序,对节点性能进行了测试和应用分析。更多还原

【Abstract】 With the development of our society and economy, the quantity of huge buildings, with more complex structure, is increasing rapidly. This brings a stricter requirement for the rescuing system in sudden disasters such as fire and blaster. Owning many advantages such as independence on infrastructure, distributed database and high reliability, wireless sensor network(WSN) can provide a different new ways for huge building disaster rescuing system. Based on traditional fire-alarm system with field bus, this paper introduces wireless sensor network into huge building field to build up a stronger and more efficient rescuing system with accurate localization ability, fast information transmission and longer lifetime. In this paper, the exploration and research of theory and application are focused on several key techniques.Firstly, the traditional fire-alarm system based on field bus is developed for large buildings. Based on advanced microcontroller, a solution with multiple protocols combination and modular design is proposed and implemented. This brings the system good scalability to adapt different requirement. At the same time, a portable information terminal for disaster rescuing is developed. It can acquire real-time information from the rescuing network and implement an efficient interaction with multimedia message between the rescuing personel and control center. Thus, the rescuing personel is able to build a fluent communication with their commander. This is helpful to improve the rescuing efficiency.Secondly, this paper surveys on the methods for fire point localization and multi-target localization with wireless sensor network. Based on the Gaussian space multi-peek fire model, the multi-source localization algorithm is improved to generate a new EM-MR method. Then, Tabu search algorithm is used for parameters estimation and acquires excellent result. The proposed method has improved accuracy of the localization effectively and decreased the computational complexity. A model for acoustic source localization is also built up. Improved MR method and Tabu algorithms are applied into multiple acoustic sources localization and reveal good performance. Simulation result verified the effectiveness of the proposed algorithms.Thirdly, the determistic deployment of wireless sensor network in building space is researched in areas with obstacles. Sensor’s detection models and coverage quality are set up. Based on the probabilistic detection model with false alarm rate, this paper proposes a new deployment method. Watershed algorithm is employed to choose the deploying sub-area. Then Delaunay Triangulation is used to generate the candidate positions for new nodes. With the evaluation mechanism, the placement of WSN nodes is realized orderly and efficiently. Contrast with different methods such as random deployment, MAX_MIN_COV and MAX_AVG_COV, the proposed method attains higher detection probability and better coverage uniformity.Due to the strict requirement of fast and accurate information transmission during building disasters, this paper adapts a data fusion tree in cluster. Furthermore, we creates a fusion tree with the smallest transmitting nodes and the shortest path based on the apex coverage technology. Then, the nodes’fusion time is assigned according to the tree structure to decrease the transmission delay and conflicts. With the Graph Theory and clustering technology, a distributed heterogeneous data fusion C-means algorithm is proposed based on clustering protocol and apex coverage for data in clusters. Simulation experiment is implemented and verifies the algorithm’s effectiveness.Finally, a new wireless sensor network node based on Zigbee is designed because current available nodes always show technical secret-keeping and can not be used for routing protocol and specific signal detection. On analysis of the Zigbee protocol struction and the features of different layers, a micro-controller and a Zigbee chip is used to form the node’s hardware. TinyOS from Micaz platform is ported into the new node and works for the modular function. An application for star-network of multiple WSN nodes is carried out. Experiment and application ana is done to analyze the WSN node’s performance.更多还原