融合地磁与RSSI的室内定位粒子滤波改进算法研究

发布时间：2018-07-27 12:00

【摘要】：随着移动互联网技术的迅速发展,基于位置的服务(Location Based Service,LBS)得到人们更多的关注。在室外环境中,蜂窝网络技术配合全球定位系统(Global Positioning System,GPS)可以实现快速准确的定位效果,已经广泛应用在各行业中。而现代社会中,调查表明普通人平均每天至少有80%的时间处于室内环境,如何提供室内环境下的便捷定位服务成为亟待解决的问题。相关研究人员已经提出过初见成效的室内定位方案:蓝牙、超宽带、射频、WiFi信号及光跟踪等,但是不同技术的定位性能都需要外在条件支持。比如,无线定位技术需要部署成本较高的额外设施,同时在信号传播时会受到室内复杂因素的干扰,产生反射,折射、多路径效应等,无法保证高精度。因此,单一的定位方案并不能很好地满足大众对于各种室内场景的定位服务要求。近年来,研究人员发现现代建筑物内部钢筋混凝土结构在局部范围产生地磁异常场,其特征信息长时间内保持稳定。理论上,可利用不同位置的磁场特征矢量唯一性进行室内定位,但在大范围内实现室内地磁匹配时,会出现初始滤波收敛缓慢,相似磁场区域造成定位漂移等问题。另一方面,在无线通信技术快速发展的推动下,大型室内场所遍布WiFi热点。如果采用基于传统信号传播模型的定位技术,研究者可获得较高的定位精度,但硬件设施昂贵,仅适合理想条件下室内场景。而基于WiFi信号强度(RSSI)的位置指纹算法适用性强,操作简便,但定位精度较低。充分考虑到地磁与WiFi定位方案的优缺点,本文提出一种融合室内地磁和WiFi信号强度的定位算法,在大范围定位区域中先通过无线信号强度粗定位到小区域,再通过地磁滤波进行下一步精匹配。本文先阐述地磁场特点和室内地磁匹配原理,然后介绍了几种主流室内定位技术原理和方法。接下来通过实验总结出钢筋混凝土结构建筑物内部的磁场环境特点。随后为了减少磁场测量误差,提出一种改进限幅滤波与均值滤波算法相结合的思路,对原始观测数据进行预处理。又对磁传感器进行硬铁补偿来减少测量噪声影响。通过实验证明了克里金插值算法在室内局部地磁图构建中具有较高精度。本文提出的融合地磁与RSSI的定位算法共分为数据建库和匹配定位两个阶段。建库阶段,提前收集定位区域的磁场强度和RSSI观测值并建立数据库。定位阶段,先通过RSSI观测指纹,采用所提加分法判定出目标所在小区域,随后进行下一步的地磁匹配。融合算法缩小了地磁匹配中随机粒子的初始搜索范围,减少粒子数量和收敛时间,提高系统定位效率。对于经典地磁粒子滤波算法匹配中出现的目标估计位置漂移现象,提出利用误差圆约束思路,来避免地磁图中相似野值点带来的干扰,从而提高定位成功率。将改进后的融合算法与经典地磁滤波算法进行实验对比,结果证明,本文算法有效地提高系统定位精度、降低匹配中滤波收敛时间,增强系统定位稳定性,可作为一种室内定位方案加以发展。
[Abstract]:With the rapid development of mobile Internet technology, Location Based Service (LBS) has received more attention. In outdoor environment, the cellular network technology combined with the global positioning system (Global Positioning System, GPS) can achieve rapid and accurate positioning effect, and has been widely used in various industries. In the meeting, the survey showed that average people were in the indoor environment at least 80% time per day. How to provide convenient location services in the indoor environment has become an urgent problem. The relevant researchers have put forward the initial results of indoor positioning schemes: Bluetooth, UWB, radio frequency, WiFi signal and light tracking, but different technology The bit performance needs external support. For example, the wireless positioning technology needs to deploy high cost additional facilities, while the signal transmission will be disturbed by the complex factors in the room, produce reflection, refraction, multipath effect and so on, which can not guarantee high precision. Therefore, a single positioning scheme can not satisfy the public for all kinds of rooms. In recent years, researchers have found that the structure of the reinforced concrete structure inside the modern building produces geomagnetic anomaly in a local range, and its characteristic information remains stable for a long time. In theory, the interior location of the magnetic field feature vectors in different positions can be used, but the indoor geomagnetic match is realized in a large range. When matching, the initial filtering convergence is slow and the similar magnetic field region causes the positioning drift. On the other hand, the large indoor places are all over the WiFi hot spots under the rapid development of wireless communication technology. If the positioning technology based on the traditional signal propagation model is adopted, the researchers can obtain higher positioning accuracy, but the hardware is set up. The location fingerprint algorithm based on WiFi signal intensity (RSSI) is very applicable and easy to operate, but the positioning accuracy is low. Taking full account of the advantages and disadvantages of geomagnetic and WiFi positioning schemes, this paper proposes a localization algorithm for integrating indoor geomagnetic and WiFi signal intensity, which is first connected in a large range location area. In this paper, the characteristics of the geomagnetic field and the principle of indoor geomagnetic matching are introduced. Then the principles and methods of several mainstream indoor positioning techniques are introduced. The following experiments are made to summarize the characteristics of the magnetic field in the structure of the reinforced concrete structure. In order to reduce the error of magnetic field measurement, a method of improving the combination of the limiting filter and the mean filtering algorithm is proposed. The original observation data is pretreated. The magnetic sensor is compensated by hard iron to reduce the influence of measurement noise. It is proved that the Kriging interpolation algorithm has a high precision in the construction of the local geomagnetic map. The fusion geomagnetism and RSSI localization algorithm proposed in this paper is divided into two stages: data building and matching location. In the building phase, the magnetic field intensity and the RSSI observation value of the location area are collected and the database is set up in advance. The location stage is first through the RSSI observation fingerprint, and the proposed addition method is used to determine the small area of the target, and then the next step is carried out. The fusion algorithm reduces the initial search range of the random particles in the geomagnetic matching, reduces the number of particles and the time of convergence, and improves the system location efficiency. The error circle constraint method is proposed to avoid the similar field values in the geomagnetic map. The results of the improved fusion algorithm are compared with the classical geomagnetic filtering algorithm. The results show that the algorithm can effectively improve the positioning accuracy of the system, reduce the convergence time in the matching and enhance the stability of the system, and can be developed as an indoor positioning scheme.
【学位授予单位】：北京建筑大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN713;TN92

【参考文献】