煤矿井下无线传感器网络的RSSI定位算法研究与实现

发布时间：2018-06-25 05:56

本文选题：井下定位 + ZigBee　；参考：《内蒙古科技大学》2014年硕士论文

【摘要】：我国煤矿储备丰富，是世界上煤矿生产第一大国。随着普适计算、传感技术、大数据和无线技术的发展，物联网即将对各个行业带来较大变革。无线传感器网络是物联网重要组成部分，将其应用在“感知煤矿”会对整个煤矿生产带来深远影响。煤矿井下目标定位作为井下安监系统的一部分，弥补了卫星定位无法为井下环境提供位置信息的局限，为安全生产带来了重要保障。由于目前我国井下定位系统大多仍采用射频识别系统，不能对人员目标进行及时定位，因此，通过对无线传感器网络、矿井定位系统及定位算法研究现状进行研究分析，确定了本文研究方向——面向煤矿井下基于RSSI的无线传感器网络定位技术。首先，介绍定位算法的基本概念、原理及分类，具体研究了几种经典定位算法。其中，参数估计定位算法主要特点在于信号的参数容易从硬件上获得，采集样本工作量小，然而在室内、井下等复杂环境中参数极易受到影响，导致测距不准。指纹匹配算法把这些受到环境影响的信息作为指纹信息的一部分，利用模式匹配相关算法在恶劣环境中能够获得更高的定位精度。其次，本文针对煤矿井下信道环境，提出了一种基于虚拟数据集和马尔科夫链的指纹匹配定位算法。在离线阶段，利用卡尔曼滤波原理获得更加稳定的数据样本。针对指纹匹配算法数据采集工作量大的不足，提出构建虚拟数据集的方案。文中对指纹匹配算法进行了详细的介绍，采用基于贝叶斯准则框架的概率型算法，它考虑到每处采样点RSSI分布先验假设和统计特征，其中核函数法将每处采样点上的每个样本数据赋予一个以自身为“核心”的函数来构建概率密度分布，避免采用确定模型带来的误差。为优化定位结果，考虑到先验概率对贝叶斯后验概率的影响，提出了基于高斯模型的马尔科夫链定位算法，并对基于粒子滤波的实时追踪算法进行仿真。最后，采用TI公司的Z-Stack协议栈和IOT-2530节点对ZigBee井下定位网络进行搭建，离线数据经Matlab处理存储在MySQL数据库中。实验表明采用经卡尔曼滤波后的数据集定位精度有所提高；利用经参数估计法建立的虚拟数据集，可以大大降低离线采集阶段的工作量；在线阶段，采用该数据集结合基于高斯模型的马尔科夫链算法在长为80m的区域内进行实验，定位均方根误差为4.52m，，满足井下目标定位所需。
[Abstract]:With the development of pervasive computing , sensing technology , large data and wireless technology , the Internet of Things is about to bring great change to each industry . Wireless sensor network is an important part of Internet of Things , and its application in " perception coal mine " will have far - reaching impact on the whole coal mine production .

As part of the underground safety monitoring system , the target location of underground coal mine is used as part of the underground safety monitoring system , which makes up the limitation of the location information of the underground environment and the important guarantee for the safety production . Because of the fact that the underground positioning system in China still adopts the radio frequency identification system , the personnel target cannot be positioned in time , the research direction of the paper is determined , which is the wireless sensor network positioning technology based on RSSI in the coal mine .

Firstly , the basic concept , principle and classification of location algorithm are introduced , and several classical positioning algorithms are studied .

Secondly , aiming at the underground channel environment of coal mine , a fingerprint matching location algorithm based on virtual data set and Markov chain is proposed . In the off - line phase , a more stable data sample is obtained by using Kalman filter principle .

In the end , using the Z - Stack protocol stack of TI Company and the node to set up the ZigBee down - hole location network , the offline data is stored in MySQL database via Matlab . The experiment shows that the positioning accuracy of the data set after Kalman filter is improved ;
By using the virtual data set established by the parameter estimation method , the workload of the off - line acquisition phase can be greatly reduced ;
In the online phase , the Markov chain algorithm based on Gaussian model is used to carry out the experiment in the area of 80 m in length , and the root mean square error is 4.52m , which can meet the requirement of downhole target positioning .
【学位授予单位】：内蒙古科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN929.5;TP212.9

【参考文献】