基于高斯混合无迹粒子滤波的地形辅助导航算法  

Terrain aided navigation based on Gaussian mixture unscented particle filter

作　　者：孙立国[1] 李欣[1] 李世丹[1] 王德生[1] 

机构地区：[1]清华大学电子工程系,北京100084

出　　处：《中国惯性技术学报》2011年第3期298-301,共4页Journal of Chinese Inertial Technology

基　　金：国家自然科学基金(60972022)

摘　　要：地形辅助导航是一种利用地形高度信息定位的导航技术,由于地形高度起伏是非线性的,因此地形辅助导航本质是非线性、非高斯贝叶斯后验概率估计问题。粒子滤波因为适合非线性、非高斯估计问题,被引入地形辅助导航领域得到广泛研究和应用,但粒子滤波算法存在粒子匮乏的问题,会影响定位精度。针对此问题,将高斯混合无迹粒子滤波(GMUPF)用于地形辅助导航,该算法用高斯混合模型(GMM)近似粒子分布,用无迹卡尔曼滤波(UKF)估计重要密度函数,不需要做重采样。通过用实际地形数据做飞行仿真实验,结果显示相比粒子滤波,不仅没有粒子匮乏问题,而且所用粒子数更少时估计精度略好。Terrain aided navigation technology uses terrain high information to estimate the aircraft position.Since the terrain height variations are nonlinear,the terrain aided navigation uses non-linear and non-Gaussian Bayesian posterior probability estimation in essential.Particle filter has been widely studied and applied in the field of terrain aided navigation,but the particle degeneracy of particle filter will affect the positioning accuracy.In this paper,Gaussian mixed unscented particle filter（GMUPF） is used in the terrain aided navigation algorithm because of the particle degeneracy of particle filter.In the algorithm,Gaussian mixture model（GMM） is used to approximate the particles distribution,and the unscented Kalman filter is used to estimate the important density function,thus resampling is not needed.The algorithm is simulated under real terrain data,and the result shows that the GMUPF can give more accurate estimation though using less particles.

关 键 词：地形辅助导航 贝叶斯后验概率估计 粒子滤波 高斯混合无迹粒子滤波 

分 类 号：U666.1[交通运输工程—船舶及航道工程]