基于微机电惯性测量的行人导航技术研究

发布时间：2018-05-13 15:23

本文选题：捷联式惯性导航 + 误差标定　；参考：《哈尔滨工业大学》2017年硕士论文

【摘要】：随着卫星定位技术的民用化和无线互联网技术的快速发展,定位导航应用已经融入到日常生活的方方面面。由于人们对室内定位需求的不断增加,如在商场、展会等室内场所中通常需要了解人员的位置信息。室内无线定位技术需要在室内环境中布设大量的无线信号发射器,价格不菲且定位精度有待提高。惯性导航技术仅依靠自身传感器实现自主导航为行人导航提供了技术参考。目前已有相应的产品应用到特殊的领域中,如消防员在室内火灾现场的定位和矿井下工人的定位等,但定位设备的惯性传感器精度较高且造价成本高,不适用于消费电子产品。因此,本文基于低成本的惯性传感器对行人导航技术展开研究。论文首先对捷联式惯性导航理论进行研究,探讨适合用于行人导航的更新算法。导航姿态矩阵计算是捷联惯性导航系统中关键算法,构建载体坐标系到导航坐标系的变换关系模型,用四元数法建立姿态更新的微分方程,并通过龙格-库塔法求解方程。再将测量加速度通过姿态矩阵转换到导航坐标系下,利用积分算法求得速度和位置等导航参数。由于低成本的MEMS三轴矢量场传感器(陀螺仪、加速度计和磁力计)存在较严重的输出误差,本文针对传感器误差的标定与补偿方法进行探讨与研究。三轴矢量场传感器由于制造工艺的差异存在零偏、标度因数与不正交角误差。本文对传感器误差建立数学模型,采用最小二乘椭球拟合的方法来确定误差参数大小。通过实验验证该方法能有效估算传感器的主要误差参数,经过误差补偿后可改善导航系统的精度。由于传感器随机漂移误差难以在使用前进行标定,需结合滤波器以减小误差的累积。互补滤波器结构简单、计算量小,与梯度下降法结合能使误差快速收敛;扩展卡尔曼滤波通过建立状态方程和量测方程能有效估计陀螺仪的随机漂移量。本文提出根据观测量可信时刻建立相应的量测方程,能有效提高估计精度。经过滤波器对惯性传感器数据融合后仍存在较严重的速度漂移,根据行人行走的步态特征通过零速判断和修正的方法校正。最后,本文对这两种滤波算法进行对比实验,以分析其应用于行人定位导航的精确度情况。第一项实验在旋转平台上进行,转台能提供精确的旋转角度信息,通过导航系统输出的姿态信息与转台给定角度作比较,分析传感器数据融合的实时性和跟随精度。第二项实验是在室内环境中检验行人导航系统的定位轨迹并分析误差。两项实验结果表明本文设计的传感器数据融合算法能很好满足行人导航的需求。
[Abstract]:With the civilian use of satellite positioning technology and the rapid development of wireless Internet technology, positioning and navigation applications have been integrated into every aspect of daily life. Due to the increasing demand for indoor positioning, people usually need to know the location information in the shopping mall, exhibition and other indoor places. Indoor wireless location technology needs to set up a large number of wireless signal transmitters in indoor environment, which is expensive and needs to be improved. Inertial navigation technology only rely on its own sensors to achieve autonomous navigation for pedestrian navigation provides a technical reference. At present, the corresponding products have been applied to special fields, such as firemen's location in indoor fire scene and mine workers' positioning, but the inertial sensor of positioning equipment has high precision and high cost, so it is not suitable for consumer electronic products. Therefore, based on the low-cost inertial sensor, pedestrian navigation technology is studied in this paper. In this paper, the theory of strapdown inertial navigation is studied, and the updating algorithm for pedestrian navigation is discussed. The calculation of navigation attitude matrix is a key algorithm in strapdown inertial navigation system. The transformation relationship model from carrier coordinate system to navigation coordinate system is constructed, and the differential equation of attitude updating is established by quaternion method, and the equation is solved by Runge-Kutta method. Then the acceleration measurement is transformed into the navigation coordinate system by attitude matrix, and the navigation parameters such as velocity and position are obtained by integral algorithm. Due to the serious output errors of MEMS three-axis vector field sensors (gyroscopes accelerometers and magnetometers) with low cost the calibration and compensation methods of sensor errors are discussed and studied in this paper. Three axis vector field sensor has zero deviation and scale factor error due to the difference of manufacturing process. In this paper, the mathematical model of sensor error is established, and the error parameter is determined by least square ellipsoid fitting method. The experimental results show that the method can effectively estimate the main error parameters of the sensor and improve the accuracy of the navigation system after error compensation. Because the random drift error of the sensor is difficult to calibrate before use, it is necessary to combine the filter to reduce the accumulation of errors. The complementary filter has the advantages of simple structure and small computational complexity, and the extended Kalman filter can effectively estimate the random drift of gyroscope by establishing the state equation and measuring equation, and combining with gradient descent method can make the error converge rapidly. In this paper, it is proposed that the estimation accuracy can be improved effectively by establishing the corresponding measurement equation according to the credible time of the observation. After the data fusion of the inertial sensor by the filter, there is still a serious velocity drift, which is corrected by the method of zero speed judgment and correction according to the gait characteristics of the pedestrian. Finally, this paper compares the two filtering algorithms to analyze the accuracy of their application in pedestrian navigation. The first experiment is carried out on the rotating platform. The turntable can provide accurate rotation angle information. The real-time and follow accuracy of sensor data fusion are analyzed by comparing the attitude information output by the navigation system with the given angle of the turntable. The second experiment is to test the location track of pedestrian navigation system and analyze the error in indoor environment. Two experimental results show that the sensor data fusion algorithm designed in this paper can well meet the needs of pedestrian navigation.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP212.9

【参考文献】