煤矿井下基于网格划分的分层定位算法研究

发布时间：2018-12-31 19:10

【摘要】：矿山物联网技术的发展推动了Wi Fi技术在煤矿井下的应用。如何在Wi Fi网络上实现定位功能,用以实现矿工位置信息的跟踪成为目前研究热点之一。针对现有地面室内定位技术在煤矿井下定位效果不佳。本学位论文对基于Wi Fi网络的煤矿井下定位进行了研究。本文首先对基于Wi Fi网络的定位算法进行了简介,在对各种算法的适用性比较的基础上,指出煤矿井下定位应采用基于场景的定位算法。传统的场景定位算法利用指纹匹配的思想进行定位,本文利用煤矿井下场景定位的特点,将统计机器学习理论引入到定位中,使用支持向量机,将场景定位中的指纹匹配问题转换成支持向量机中的分类问题。针对分类的准确性问题。对如何优化支持向量机参数进行了研究,并利用启发式算法优化对支持向量机参数进行优化。通过仿真分析看出,通过启发式算法优化后的支持向量机,分类准确度最高可以达到98.88%。在对井下实际定位场景环境分析的基础上,本文提出了基于网格划分的分层定位算法。算法实现定位从大范围到小区域的逐步精化。该算法与传统场景定位算法相比,充分发挥了传统场景算法优势,又有效避开了传统算法的不足,实验结果说明该算法可以获得更好定位精度和稳定性,与常用的定位系统相比算法平均定位精度提高约10%。
[Abstract]:The development of mine Internet of things technology has promoted the application of Wi Fi technology in coal mine. How to realize the location function on Wi Fi network and how to track the miners' position information has become one of the research hotspots. In view of the existing surface indoor positioning technology in coal mine underground positioning effect is not good. In this thesis, the location of underground coal mine based on Wi Fi network is studied. In this paper, the localization algorithm based on Wi Fi network is introduced. Based on the comparison of the applicability of various algorithms, it is pointed out that the location algorithm should be based on scene. The traditional scene location algorithm uses the idea of fingerprint matching to locate. In this paper, the statistical machine learning theory is introduced into the location, and the support vector machine is used. The fingerprint matching problem in scene location is transformed into a classification problem in support vector machine (SVM). Aim at the accuracy of classification. This paper studies how to optimize the parameters of support vector machine, and uses heuristic algorithm to optimize the parameters of support vector machine. The simulation results show that the classification accuracy can reach 98.88 by the optimized support vector machine based on heuristic algorithm. Based on the analysis of the environment of the downhole actual location scene, this paper presents a hierarchical localization algorithm based on grid division. The algorithm realizes the gradual refinement of localization from a large area to a small area. Compared with the traditional scene location algorithm, this algorithm has the advantage of the traditional scene algorithm, and effectively avoids the shortcomings of the traditional algorithm. The experimental results show that the algorithm can achieve better positioning accuracy and stability. Compared with the common positioning system, the average positioning accuracy of the algorithm is improved by about 10%.
【学位授予单位】：中国矿业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TD65

【参考文献】