一种SINS/GPS紧组合导航系统的改进自适应扩展卡尔曼滤波算法

针对自适应扩展卡尔曼滤波算法中系统噪声协方差矩阵与量测噪声协方差矩阵不能同时被估计的问题，提出了一种改进的自适应扩展卡尔曼滤波算法.该算法基于残差，主要对滤波算法中的自适应估计器进行改进，改进后可以实时估计系统噪声.基于该算法，设计了新的滤波器并应用在SINS/GPS紧组合导航系统上，可随着系统中噪声的变化而自动地调节协方差矩阵.最后，分别用扩展卡尔曼滤波和改进的自适应扩展卡尔曼滤波对SINS/GPS紧组合模型进行仿真，结果表明改进的自适应的扩展卡尔曼滤波比扩展卡尔曼滤波的定位误差与测速误差更小，滤波的稳定性更好. 

An Improved Adaptive Extended Kalman Filtering Algorithm of SINS/GPS Tightly-Coupled Integrated Navigation System

To deal with the problem that process noise covariance matrix and measurement noise covariance matrix in adaptive extended Kalman filtering algorithm cannot be estimated at the same time, a new kind of improved adaptive extended Kalman filtering algorithm was proposed. Based on residual sequence, this algorithm mainly improved the adaptive estimator of filtering algorithm, which could estimate process noise at real-time after improvement. Based on this algorithm, a new filter was designed to be applied to SINS/GPS tightly-coupled integrated navigation system, which could automatically adjust covariance matrix as noise varied in the system. Finally, extended Kalman filtering (EKF) and the improved adaptive extended Kalman filtering (AEKF) were applied respectively to simulate SINS/GPS tightly-coupled models. Tests show that the improved adaptive extended Kalman filtering has fewer positioning errors and velocity errors, and better stability of filtering than EKF.