基于嗅觉移动传感器网络的气体源定位

发布时间：2019-06-15 04:10

【摘要】：移动传感器网络是由无线传感器网络节点配以移动平台组合而成，故而移动传感器网络在具有网络特性的同时增加了机动性，这有效地扩大了传感器节点的作用范围，通过少数节点的不断移动就有望实现部署大量静止节点达到的效果。移动传感器网络在有毒/有害气体泄露检测、火源探测和灾后搜救等方面具有广阔的应用前景。本文围绕移动传感器网络及其在气体泄漏源定位中的应用问题，重点开展了如下几个方面的研究工作。首先，根据移动传感器网络对节点的体积小、价格便宜等实际要求，设计了适用于气体源定位的移动节点。并针对此移动传感器网络，给出了节点的运动控制方法，具有自恢复功能的网络协议以及可抑制零漂的目标传感器标定措施。其次，给出了一种分布式的基于声音的移动节点相对定位算法，并通过理论和实验证明：在声音传感范围内的多个节点，只要有两个节点相继发声，即可实现所有节点之间的相互定位。所提算法的方向估计的过程中对卷积的加窗运算有效地减小了计算量；同时声源距离估计中采集信号之间的代数运算有效地降低了环境中的随机噪声。再次，提出了一种使用固定拓扑的移动传感器网络对气体源进行定位的方法，所提定位算法中移动节点不需对风信息进行实时测量。该算法可以看作是一个通过有限移动节点的传感器网络的动态部署实现对气体源进行估计的过程。在时均风恒定的环境下，每个估计周期分两步，首先对环境中的气体进行一段时间的测量，根据测量的浓度均值使用最小二乘法对气体源位置参数进行估计；接着在朝向估计值的方向上选择一点作为网络的几何中心，利用饱和控制的方法使节点再重新部署网络。另外，，在风随机变化的环境但没有风传感器的情况下，基于固定拓扑的移动节点采集得到的烟羽结构信息，提出了一种烟羽跟踪方法。所提方法的有效性通过仿真及实验得到了验证。最后，提出了一种基于条件信息熵，使用网络连接受限的移动传感器网络进行气体源定位算法。该算法中使用分布式粒子滤波近似计算条件信息熵及其梯度。并使用条件信息熵梯度构建了移动节点的控制律，控制节点沿条件信息熵负梯度方向运动，从而减小条件信息熵，增加对气体源位置估计的确定性。移动节点使用该算法可以实现在风不断变化或者存在障碍物的环境下的气体源定位。不同环境的仿真表明了该算法的有效性。
[Abstract]:The mobile sensor network is composed of wireless sensor network nodes and mobile platform, so the mobile sensor network not only has the network characteristics, but also increases the maneuverability, which effectively expands the scope of the sensor nodes. Through the continuous movement of a small number of nodes, it is expected to achieve the effect of deploying a large number of static nodes. Mobile sensor networks have broad application prospects in toxic / harmful gas leakage detection, fire source detection and post-disaster search and rescue. In this paper, focusing on the mobile sensor network and its application in gas leakage source location, the following research work has been carried out. First of all, according to the actual requirements of mobile sensor networks, such as small size and low price, a mobile node suitable for gas source location is designed. Aiming at this mobile sensor network, the motion control method of the node, the network protocol with self-recovery function and the calibration measure of the target sensor which can restrain the zero drift are given. Secondly, a distributed relative localization algorithm of mobile nodes based on sound is proposed, and it is proved by theory and experiment that as long as there are two nodes in the range of sound sensing, the mutual location between all nodes can be realized. In the process of direction estimation of the proposed algorithm, the windowing operation of convolution effectively reduces the amount of computation, and the algebra operation between the collected signals in the estimation of sound source distance effectively reduces the random noise in the environment. Thirdly, a fixed topology mobile sensor network is proposed to locate the gas source. In the proposed location algorithm, the mobile node does not need to measure the wind information in real time. The algorithm can be regarded as a process of gas source estimation through the dynamic deployment of sensor networks with limited mobile nodes. In the environment with constant time-averaged wind, each estimation period is divided into two steps. Firstly, the gas in the environment is measured for a period of time, and the position parameters of the gas source are estimated by the least square method according to the measured concentration mean value. Then, a point is selected as the geometric center of the network in the direction of the estimated value, and the node is redeployed by the method of saturation control. In addition, in the case of randomly changing wind environment but no wind sensor, a plume tracking method is proposed based on the fixed topology of the smoke structure information collected by the mobile node. The effectiveness of the proposed method is verified by simulation and experiments. Finally, a gas source location algorithm based on conditional information entropy and limited network connection is proposed. In this algorithm, distributed particle filtering is used to approximate the conditional information entropy and its gradient. The control law of the mobile node is constructed by using the conditional information entropy gradient, and the control node moves along the negative gradient direction of the conditional information entropy, thus reducing the conditional information entropy and increasing the certainty of the position estimation of the gas source. The mobile node can use this algorithm to locate the gas source in the environment where the wind is constantly changing or there are obstacles. The simulation results in different environments show the effectiveness of the algorithm.
【学位授予单位】：天津大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TP212.9;TN929.5

【参考文献】