基于地磁场和智能手机的粒子滤波室内定位算法
发布时间：2018-12-10 12:16
【摘要】：随着计算机技术和人工智能的快速发展,基于室内定位服务的应用不断增多。其中,基于地磁场的室内定位技术因为无需外在设施、定位精度高等独特的优点逐渐成为研究热点,粒子滤波算法被认为是基于地磁场的各种室内定位算法中最有前景的算法之一。但是,现有的基于粒子滤波的室内定位算法一方面普遍存在着重采样之后粒子贫乏的问题,另一方面又受到行为模型的误差影响而定位误差大,系统可靠性不高。本文针对粒子滤波定位算法中存在的问题进行分析与研究,通过在重采样算法、观测模型以及行为模型等多个环节进行改进,提出了一个改进的粒子滤波算法,并且通过仿真实验证明本文提出的定位算法误差在1米左右。本文主要工作是:通过分析智能手机传感器的测量特性,在利用不同手机测量构建地磁场模型的实验基础上,提出以地磁变化率改进算法执行过程中对地磁数据的处理;通过分析粒子权重对重采用过程的重要性,提出基于观测路径相似性的采样方法,并采用相对误差计算相似度;针对粒子滤波算法在重采样之后粒子丰富性降低的问题,提出了一种改进的进化重采样粒子滤波算法;通过分析行为模型误差,提出了以粒子加权步长代替运动模型中的固定步长;通过分析定位目标朝向发生改变增加水平方向磁场噪声的情况,提出了一种融合三维分量匹配模型和HV匹配模型的混合匹配模型;最后针对算法定位丢失定位目标的情况,提出了一种地磁匹配的解决方法。
[Abstract]:With the rapid development of computer technology and artificial intelligence, the application of indoor positioning service is increasing. Among them, indoor positioning technology based on geomagnetic field has gradually become a research hotspot because of its unique advantages such as high positioning accuracy and no need for external facilities. Particle filter algorithm is considered as one of the most promising indoor localization algorithms based on geomagnetic field. However, the existing indoor localization algorithms based on particle filter have the problem of poor particles after resampling, on the other hand, because of the error of behavior model, the localization error is large and the system reliability is not high. In this paper, the problems in particle filter localization algorithm are analyzed and studied. An improved particle filter algorithm is proposed by improving the resampling algorithm, observation model and behavior model. The simulation results show that the error of the proposed algorithm is about 1 meter. The main work of this paper is as follows: by analyzing the measurement characteristics of smart phone sensors and on the basis of the experiment of using different mobile phone measurements to construct geomagnetic field model, this paper puts forward the processing of geomagnetic data in the execution process of the improved algorithm of geomagnetic variation rate; By analyzing the importance of particle weight to the re-adoption process, a sampling method based on the similarity of observation path is proposed, and the similarity is calculated by using relative error. An improved evolutionary resampling particle filter algorithm is proposed to solve the problem of decreasing particle richness after resampling. By analyzing the behavior model error, the particle weighted step size is proposed to replace the fixed step size in the motion model. By analyzing the situation that the orientation of the target is changed to increase the horizontal magnetic field noise, a hybrid matching model is proposed, which combines the 3D component matching model and the HV matching model. Finally, a method of geomagnetic matching is proposed for locating lost targets.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN713
本文编号：2370550
[Abstract]:With the rapid development of computer technology and artificial intelligence, the application of indoor positioning service is increasing. Among them, indoor positioning technology based on geomagnetic field has gradually become a research hotspot because of its unique advantages such as high positioning accuracy and no need for external facilities. Particle filter algorithm is considered as one of the most promising indoor localization algorithms based on geomagnetic field. However, the existing indoor localization algorithms based on particle filter have the problem of poor particles after resampling, on the other hand, because of the error of behavior model, the localization error is large and the system reliability is not high. In this paper, the problems in particle filter localization algorithm are analyzed and studied. An improved particle filter algorithm is proposed by improving the resampling algorithm, observation model and behavior model. The simulation results show that the error of the proposed algorithm is about 1 meter. The main work of this paper is as follows: by analyzing the measurement characteristics of smart phone sensors and on the basis of the experiment of using different mobile phone measurements to construct geomagnetic field model, this paper puts forward the processing of geomagnetic data in the execution process of the improved algorithm of geomagnetic variation rate; By analyzing the importance of particle weight to the re-adoption process, a sampling method based on the similarity of observation path is proposed, and the similarity is calculated by using relative error. An improved evolutionary resampling particle filter algorithm is proposed to solve the problem of decreasing particle richness after resampling. By analyzing the behavior model error, the particle weighted step size is proposed to replace the fixed step size in the motion model. By analyzing the situation that the orientation of the target is changed to increase the horizontal magnetic field noise, a hybrid matching model is proposed, which combines the 3D component matching model and the HV matching model. Finally, a method of geomagnetic matching is proposed for locating lost targets.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN713
【参考文献】
相关期刊论文 前7条
1 张聪聪;王新珩;董育宁;;基于地磁场的室内定位和地图构建[J];仪器仪表学报;2015年01期
2 邓翠婷;黄朝艳;赵华;田海冬;;地磁匹配导航算法综述[J];科学技术与工程;2012年24期
3 赵建虎;王胜平;王爱学;;一种改进型TERCOM水下地磁匹配导航算法[J];武汉大学学报(信息科学版);2009年11期
4 叶龙;王京玲;张勤;;遗传重采样粒子滤波器[J];自动化学报;2007年08期
5 余洪山;王耀南;;基于粒子滤波器的移动机器人定位和地图创建研究进展[J];机器人;2007年03期
6 房建成;宁晓琳;;月球探测器天文导航的遗传粒子滤波方法[J];北京航空航天大学学报;2006年11期
7 包志军,马培荪,仝建刚,王春雨;人行走速度规律的实验研究[J];实验室研究与探索;2000年06期
相关博士学位论文 前2条
1 赵立军;室内服务机器人移动定位技术研究[D];哈尔滨工业大学;2009年
2 殷波;移动机器人同时定位与地图创建方法研究[D];中国海洋大学;2006年
相关硕士学位论文 前1条
1 丁家栋;基于Rao-Blackwellized粒子滤波的移动机器人同时定位与地图创建算法的研究[D];合肥工业大学;2013年
，本文编号：2370550
本文链接：https://www.wllwen.com/kejilunwen/dianzigongchenglunwen/2370550.html
