卡尔曼滤波在GPS定位中的研究与实现

发布时间：2018-06-16 08:53

本文选题：GPS + EKF　；参考：《电子科技大学》2013年硕士论文

【摘要】：本文以全球定位系统（GPS，Global Positioning System）为研究背景，在利用GPS伪距测量值进行绝对定位的解算基础之上，为了提高解算的精度，，研究了当前在GPS导航中常用的几种运动模型和卡尔曼滤波（KF，Kalman Filter）算法，详细介绍了各种滤波算法的原理，实现过程，对比各自的性能优劣，为实际运用提供有效的理论指导。首先本文研究分析了几种常用的机动目标方程，分为单模型和交互式多模型（IMM，Interacting Muliple Model）。重点研究了单模型中的辛格（Singer）模型，当前统计（CS，Current Statistical）模型，分析对比了这两种模型在不同的运动状态下的估计精度，为实际的工程应用提供有效的依据。此外由于传统的CS模型中，加速度方差的不合理取值，在不增加算法复杂度的情况下，提出了两种简单可行的改进模型，通过加速度方差自适应选取合适的值，实时地调整模型参数，使得模型更加地与实际相符。由于卡尔曼滤波（KF，Kalman Filter）可以通过物体的运动方程去将用户相邻时刻的运动状态信息联系起来，使得解算结果更加的平滑。因此在GPS滤波算法这一部分，本文重点研究了目前在GPS中常用的，适用于非线性系统的扩展卡尔曼滤波（EKF，Extend Kalman Filter）方法，改进的无迹卡尔曼滤波（UKF，Unscented Kalman Filter）算法，以及用于保证误差协方差矩阵非负定性和对称性的平方根无迹卡尔曼滤波（SRUKF，Square Root Unscented Kalman Filter）算法，分析对比了各种算法的性能优劣以及适用条件，为实际的工程应用提供有效的依据。最后本文在TI OMAP3530EVM上面，制作了QTE界面，完成以上算法在OMAP3530EVM上的开发，并且对仿真结果进行验证。为实际应用中，运动状态方程以及滤波算法的选择提供参考。
[Abstract]:In this paper, the GPS GPS Global Positioning system is used as the research background, on the basis of using the GPS pseudo-range measurement value to solve the absolute positioning, in order to improve the accuracy of the solution, Several motion models and Kalman filter algorithms commonly used in GPS navigation are studied in this paper. The principle and implementation process of various filtering algorithms are introduced in detail, and their performance is compared to provide effective theoretical guidance for practical application. In this paper, several commonly used maneuvering target equations are studied and analyzed, which can be divided into single model and interactive multiple model. The Singer model in single model and the current statistical model in current statistical model are studied. The estimation accuracy of these two models under different motion states is analyzed and compared, which provides an effective basis for practical engineering application. In addition, because of the unreasonable value of acceleration variance in the traditional CS model, two simple and feasible improved models are proposed without increasing the complexity of the algorithm, and the appropriate values are adaptively selected through the acceleration variance. The model parameters are adjusted in real time to make the model more consistent with the reality. Because the Kalman filter can connect the motion state information of the user at the adjacent time through the motion equation of the object, the result of the solution is smoother. Therefore, in the part of GPS filtering algorithm, this paper focuses on the extended Kalman filter (EKFO extend Kalman filter) method, an improved unscented Kalman filter algorithm, which is commonly used in GPS, which is suitable for nonlinear systems. And the square root unscented Kalman filter algorithm, which is used to guarantee the nonnegative definiteness and symmetry of error covariance matrix, is analyzed and compared, which provides an effective basis for practical engineering application. Finally, the QTE interface is made on TI OMAP3530 EVM, and the above algorithm is developed on OMAP3530 EVM, and the simulation results are verified. It provides a reference for the selection of motion state equations and filtering algorithms in practical applications.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P228.4;TN713

【参考文献】