粒子滤波算法在GPS/DR组合导航中的应用  

Application of Particle Filtering Algorithms in GPS/DR Integrated Navigation

作　　者：宫轶松[1] 归庆明[2] 李保利[1] 周宁[3] 

机构地区：[1]信息工程大学测绘学院,河南郑州450052 [2]信息工程大学理学院,河南郑州450001 [3]河南大学计算机与信息工程学院,河南开封475000

出　　处：《测绘科学技术学报》2010年第1期27-30,共4页Journal of Geomatics Science and Technology

基　　金：国家自然科学基金资助项目(40974009;40474007)

摘　　要：针对建议分布函数的选择问题,系统地分析比较了改进的粒子滤波算法。在此基础上提出了一种新的粒子滤波算法——自适应渐消扩展Kalman粒子滤波方法。该方法用渐消扩展Kalman滤波产生建议分布函数,由于参数的可在线调节性,使得系统具有更好的自适应性和鲁棒性。与用转移先验、扩展Kalman滤波、自适应扩展Kalman滤波、迭代扩展Kalman滤波以及无迹Kalman滤波产生建议分布函数的粒子滤波方法相比,自适应渐消扩展Kalman粒子滤波进一步提高了粒子滤波的精度。通过对GPS与航位推算(DR)组合导航系统GPS/DR的试验,验证了该方法的有效性。In allusion to the choice of the proposal distribution function, several improved algorithms to the particle filtering were analyzed and compared, and a new particle filtering algorithm named the adaptive fading extended Kalman particle filter was proposed. This method takes advantage of the adaptive fading extended Katman filter to generate the proposal distribution function, and can tune the parameter in line, which has better self adaptability and robustness. Compared with several improved particle filtering methods whose proposal distribution function coming from the transition prior, the extended Kalman filter, the adaptive extended Kalman filter, the iterated extended Kalman filter and the unscented Kalman filter, the adaptive fading extended Kalman particle filter improves the accuracy of the particle filtering. An experiment for the GPS/DR integrated simulation system validated the effectiveness of the new approach.

关 键 词：粒子滤波 渐消滤波 遗忘因子 扩展Kalman滤波 无迹Kalman滤波 GPS/航位推算 

分 类 号：P207[天文地球—测绘科学与技术]