【作者】 杨君；

【导师】 柴利；

【作者基本信息】 武汉科技大学 ， 控制理论与控制工程， 2016， 博士

【摘要】 水环境中污染源的探测与定位问题对水环境保护有重要的意义。论文将无线传感器网络应用于水环境监测中,对网络平台总体架构、污染源探测和定位等关键问题开展了深入研究。论文的主要工作如下:(1)提出了网络污染源探测与节点污染源探测的方法。针对静态污染源探测与动态污染源探测中存在的探测问题,采用假设检验方法和相关二值假设,构造了检验统计量与检验方法,检验了提出的污染源探测算法。检验表明:在网络污染源探测中,高检测精度与具有较大监测值的观测节点的数量相关。而在节点污染源探测中,只要参照样本足够多,较少的采样时间也能得到更为满意的检测精度。(2)提出了两种湖库中近岸排放污染源的定位方法:一种方法是边界约束最小二乘污染源定位算法,另一种是边界约束最小一乘污染源定位算法。在一般湖库水环境中,两种算法都较无边界约束情形具有较好的定位效果;在定位节点中有奇异节点时,边界约束最小一乘污染源定位算法较边界约束最小二乘污染源定位算法具有更好的定位效果。(3)给出了河流中二维污染物浓度分布的通用模型。分析了有边界和无边界下污染源的扩散模型,给出了一种考虑边界影响的二维污染物浓度分布的通用模型。提出一种有边界约束的非线性最小二乘污染源定位算法。该算法以测量值与理论值之差的平方和为目标函数,具有较好数值稳定性和抗干扰性。通过仿真讨论了测量浓度噪声、节点漂移误差和反射系数误差对污染源定位估计精度的影响,验证了所提算法的有效性。(4)建立了面向水环境监测的分布式无线传感器系统结构,搭建了该系统污染源定位实验平台,通过该平台测试了污染源的探测算法与湖库水环境中污染源定位算法。更多还原

【Abstract】 Detection and localization of pollution source in water environment are very significant for water environment protection. Focusing on key issues of applying the wireless sensor network to water environment monitoring, this thesis has done extensive research on network structure and platform realization, pollutant detection and source localization, etc. The main contributions are as follows:(1) The detection methods of both networked pollution sources and node pollution sources are proposed. Following the hypothesis test method, we adopt the related binary hypothesis, testing statistics as well as inspection techniques to solve the detection problems. Moreover, the proposed pollution detection method is examined. The results show that high testing accuracy is related to the number of nodes with observation value. While in node pollution source detection, sampling time small sampling periods can lead to good enough detection accuracy as long as there are sufficient reference samples.(2) Two methods to localize offshore emissions pollution sources are presented. one is least square algorithm with boundary constraint, and the other is least one norm algorithm with boundary constraints. Both methods are better than those with without considering the boundary constraint of the lakes, rivers and other water environments The least one norm based method achieves better detection in the situation that singular localization nodes exist.(3) A general two-dimension model of pollutant concentration distribution in rivers is proposed. The diffusion models of the pollutant in river with no boundary constraints and near shore boundary constraints are analyzed, a 2-d universal model of distribution of pollutant concentration with boundary influence is presented. A nonlinear least-square method with boundary constraint is proposed. For this method, the quadratic sum of the difference between the measurement values and the theoretical values is taken as the objective function, and the method has some good numerical stability and robustness. Moreover, the influences of the noise measurement concentration, node floating error and the reflection coefficient error on accuracy of pollutant location estimation are tested by simulations. Effectiveness of the proposed algorithm is verified.(4) A distributed wireless sensor network platform for water environment is developed. The pollution source localization experimental platform is established. Based on this platform, the effectiveness of some proposed detection and localization algorithms is verified.更多还原