【作者】 易灵芝；

【导师】 丁德馨；

【作者基本信息】 南华大学 ， 安全科学与工程， 2018， 博士

【摘要】 铀尾矿库作为核燃料循环系统中数量最庞大的放射性废物贮存场所,可能会对其周边环境产生一定的放射性污染,是长期潜在的放射性污染源。为了消除铀尾矿库对周边环境和公众安全带来的放射性污染隐患,亟需构建高效的智能监测方法和理论体系对铀尾矿库周边环境进行有效监测。针对现有的铀尾矿库周边区域放射性污染监测方法效率较低、智能化程度不高等不足,本文利用地质统计学、多源信息融合等理论,提出和构建了基于无线传感器网络、物联网等新型网络理论和方法的智能监测体系,对基于无线传感器网络、物联网的铀尾矿库周边环境放射性污染目标检测、铀尾矿库周边环境监测覆盖盲区检测,以及覆盖盲区修复等一系列关键问题进行了研究,提出了基于数据融合理论的放射性污染目标高效检测方法,设计了能量高效的可信信息覆盖盲区检测方法,解决了铀尾矿库放射性污染监测物联网多模覆盖盲区修复问题。研究主要内容和成果如下:(1)基于数据融合理论,提出了一种高效节能的铀尾矿库周边环境放射性污染目标检测方法。充分挖掘传感器节点之间的协作,实现了无线传感网分布式分簇决策融合放射性污染目标检测。每个节点根据监测数据和局部决策阈值做出局部决策,然后收集邻居节点的决策结果,基于决策融合规则做出簇级决策,最后,无线传感网络根据各节点的簇级决策融合结果做出网络级决策。实验结果表明,所提方法可逐级控制各监测数据的误报率,有效提高监测数据的可靠性和精确度。(2)基于可信信息覆盖模型,提炼了铀尾矿库放射性污染监测物联网可信信息覆盖盲区检测问题,设计了考虑节点能量消耗模型的覆盖盲区检测方法。首先,基于空间相关和相关变程将检测区域分割成一系列的重建网格;然后,基于可信信息覆盖模型对每个重建网格进行扫描和检测,判断其是否为覆盖盲区;最后,利用图像处理法提取覆盖盲区边界。实验结果表明所提出的方法均能有效的检测到覆盖盲区的具体位置和数量。(3)从局域视角研究了铀尾矿库周边区域环境监测物联网可信信息覆盖盲区的检测问题,提出了综合考虑节点的通信能力、通信半径及能量损耗的局域化检测方法。利用邻居传感器节点的协作和通信能力,传感器节点的定位和分布以及监测的放射性环境变量的空间相关性,设计了包括LCHD、LCHDRL、Random和Random RL等启发式解决方案。LCHD和LCHDRL两个方案都是局域化确定每个子区域的覆盖状态并考虑了传感器的通信能力。LCHDRL方案在可信信息覆盖盲区检测过程中,不仅考虑了传感器的剩余能量,还考虑了其剩余寿命。Random和Random RL两种方案都是在传感区域任意选择传感器进行可信信息覆盖盲区检测。在获取了每个子区域的覆盖状态后,我们通过图像处理技术提取覆盖盲区的边界。实验结果表明,所提方案能高效节能的检测到覆盖盲区的位置和数量。(4)研究了基于可信信息覆盖模型的铀尾矿库放射性污染监测物联网多模覆盖盲区修复问题,设计了修复多模可信信息覆盖盲区的移动节点派遣方法。挖掘监测环境变量的空间相关性,将可信信息覆盖盲区修复问题归约为集合划分问题,设计了集中式多模可信信息覆盖盲区修复方案(C-MCICHH)、分布式多模可信信息覆盖盲区修复算法(D-MCICHH)和随机可信信息覆盖盲区修复方案(Random)等启发式覆盖盲区修复算法。通过构造一个完全加权二分图,C-MCICHH算法将覆盖盲区修复问题转化为最大权重最大匹配问题,并利用经典的匈牙利策略求解。D-MCICHH贪婪地选择移动节点及其感测单元,根据修复贡献指标确定移动节点并通过竞争机制派遣其移动至覆盖盲区。Random方案随机选择多模移动节点,集中式地派遣节点修复覆盖盲区。实验结果证实了所提方法能有效地修复可信信息覆盖盲区,优化覆盖性能,提升网络寿命。更多还原

【Abstract】 As the largest special facility and site for storing radioactive uranium tailings in the nuclear fuel recycle system,the uranium tailings pond,which may cause potential radioactive pollution to its surrounding environment,is a major hazard and a long-term potential radioactive pollution source.In order to avoid and eliminate the radioactive pollution threat caused by uranium tailings ponds to surrounding environment,it is urgent to develop an energy efficient theoretical framework and method for intelligently monitoring the environmental radioactive pollution in the surrounding regions of uranium tailings ponds.In view of the shortcomings including low efficiency and Low intellectualization in existing radioactive pollution monitoring methods and theories,based on the theories of geostatistics and multi-source information fusion,this thesis studies a series of crucial problems on how to effectively monitor the radioactive pollution by fully exploits theadvantages of wireless sensor network and Internet of Things.This thesis mainly focuses on the issuses of WSN and Io T-based radioactive pollution target detection,coverage hole detection and coverage hole healing.We have proposed a data fusion based radiolgocial pollution target detection strategy,devised several energy-efficient confident information coverage hole detection alogrithms,and sovled the multi-modal confident information coverage hole healing problem.The main contributions of this thesis are listed as follows.(1)We propose an energy-efficient scheme for detecting radioactive pollution targets in surrounding uranium tailings pond based on data fusion theory.By fully exploiting the collaboration among sensor nodes,we can effectively detect the radioactive pollution targets and sources.The process of distributed cluster decision fusion includes the following steps.Firstly,each sensor makes a local-level decision based on its own measurement and a given local decision threshold.Secondly,the scattered sensors collect the decision results from their neighbors and make a cluster-level decision based on a K-out-of-N decision fusion rule.Finally,the base station of the network makes the network-level decision from the cluster-lever fusion results of these individual sensors.Experimental results show that the proposed strategy can effectively lower the false alarm rate of each monitoring data step by step,and significantly improve the reliability and accuracy of monitoring data.(2)We define the coverage hole detection problem in Internet of Things for uranium tailings pond environmental radioactive pollution monitoring based on the confident information coverage model(CIC),and study how to energy-efficiently detect the potential CIC holes.Firstly,the sensing field is partitioned into a series of reconstruction grids based on the spatial correlation and correlation range.Then each reconstruction grid will be scanned and detected based on the CIC model to be judged whether it is a hole.Finally,the boundary of the coverage hole will be exacted by image processing method.The experimental results show that both the proposed schemes can efficiently detect the locations and the number of the emerged coverage holes.(3)We study the localized confident information coverage hole detection(LCICHD)problem of environmental radioactive pollution monitoring for uranium tailings pond based on Internet of Things,and devise a family of localized coverage hole detection protocols taking the sensors’ communication ability,communication radius and energy consumption into consideration.By making full use of the collaboration and communication capabilities of the neighbor sensor nodes,the locations and spatial distribution of sensor nodes and the spatial correlation of the monitored radioactive physical variables,we develop four heuristic CIC holes detection schemes including the LCHD,LCHDRL,Random and Random RL.Both the LCHD and LCHDRLschemes locally determine coverage status of each subregion and take the sensor communication ability into consideration.While the LCHDRL considers not only the sensor remaining energy but also the residual lifetime during the CIC hole detection.For comparison,both the Random and Random RL schemes arbitrarily select sensors within the sensing field to detect CIC holes,and the Random RL scheme takes the sensors’ residual lifetime into consideration during the hole detection process.After acquiring the coverage status of each partitioned local subregion,the coverage hole boundary will be extracted by image processing techniques.Experimental simulations show that the proposed schemes can efficiently detect the emerged coverage holes including the locations and the number.(4)Based on the novel confident information coverage model,we provide a study on how to heal the emerged multi-modal confident information coverage holes(MCICH)in Io T for radiological pollution monitoring,and give two multi-modal mobile sensor dispatching schemas for healing the CIC coverage holes.We develop a family of heuristic schemes including the centralized C-MCICHH,the distributed D-MCICHH and Random by fully exploit the spatial correlation of monitoring environment variables and the coordination of sensor nodes,after proving its NP-completeness by reducing the MCICHH problem to the classical set partition problem.The C-MCICHH converts theMCICHH problem into the maximum weight maximum matching problem by constructing a complete weighted bipartite graph to acquire the optimal solution by utilizing the classical Hungarian strategy.The D-MCICHH greedily selects mobile nodes and sensing units according to the contribution index,and sends them to the holes through the competition mechanism.The Random scheme arbitrarily selects the multi-modal mobile Io T sensors to be dispatched to heal the existing multi-modal CIC holes in a centralized manner.Simulation results validate that the proposed solutions can effectively heal the multi-modal CIC holes,optimize the network coverage performance,improve the network operation efficiency and prolong network lifetime.更多还原