高动态下紧耦合组合导航改进UKF滤波器设计  

Design and application of improved unscented kalman filter in high-dynamic tightly integrated navigation

作　　者：张良[1] 鲍其莲[1] 

机构地区：[1]上海交通大学电子信息与电气工程学院仪器科学与工程系,上海200240

出　　处：《电子测量技术》2011年第1期90-93,共4页Electronic Measurement Technology

摘　　要：由于紧耦合导航系统模型为非线性模型,采用线性化滤波方法,线性化误差将影响组合导航的精度,高动态下影响更为明显。UKF(平淡卡尔曼滤波器)避免了非线性系统模型的线性化过程,具有较好的精度。但在高动态条件下,随着滤波迭代计算的进行,舍入误差的积累会引起均方误差矩阵失去非负定性甚至对称性,使滤波增益矩阵失去合适的加权作用从而导致滤波器的发散。提出一种UKF改进算法,将平方根滤波与UKF相结合,避免了高动态下的滤波发散问题。通过仿真比较,基于Potter平方根算法改进的UKF紧耦合导航系统在高动态情况下具有更好的精度及鲁棒性。The accuracy of integrated navigation system worsens especially in High-dynamic condition due to the Linearization of nonlinear model.The unscented kalman filter is applied to improve the accuracy using nonlinear model.Potter square root filter is introduced to improve UKF keeping the positive definite state of the covariance matrix.The simulation show that the improved UKF in tightly integrated navigation system have better performance both in accuracy and robustness in high dynamic.

关 键 词：高动态 平淡卡尔曼滤波 平方根滤波Potter算法 紧耦合 组合导航 

分 类 号：U666.1[交通运输工程—船舶及航道工程]