MIMU/GNSS紧组合精密单点定位协方差成形自适应滤波方法

针对精密单点定位应用需求,研究了MIMU/GNSS紧组合协方差成形自适应滤波方法。给出了"位臵滤波器+速度滤波器"的分布式滤波器设计方案以降低计算复杂度;推导了地理系紧组合导航系统模型,并把伪距测量不一致性偏差扩展至系统状态向量中予以估计,从而提高系统对于动态环境下伪距测量偏差抖动的适应能力。协方差成形自适应滤波算法利用Frobenius范数来衡量系统残差噪声水平建模状态与实际状态的匹配程度,并以最小化Frobenius范数作为优化指标,动态调节滤波增益,以此来提高状态估计精度、平稳性与鲁棒性。地面静态试验表明:紧组合协方差成形自适应滤波器定位误差均值与均值稳定性均优于标准卡尔曼滤波器,定位精度提高了约50%,能够提供亚米级单点定位导航服务。相较于集中式滤波器设计方案,分布式滤波器方案计算复杂度降低了63.5%。

Covariance shaping adaptive filter method for tightly-coupled GNSS precise single-point positioning inertial navigation

According to the application requirement of the precision single-point positioning,the adaptive filtering method for MIMU/GNSS tightly-coupled system based on covariance shaping is studied.The distributed filter design of the “position filter + speed filter” is given to reduce the computational complexity.The model of MIMU/GNSS tightly-coupled system in geography system is deduced,and the pseudo-range measurement inconsistency is extended into the system state vector to improve the system's ability to suppress the pseudo-range deviation jitter in dynamic environments.The covariance shaping adaptive filtering algorithm utilizes the Frobenius norm to measure the matching degree between the modeling and actuality of residual noise,and dynamically adjust the filter gain to minimize the Frobenius norm which is taken as the optimization index,in order to improve the state estimation accuracy,stability and robustness.The ground static test shows that the positioning error's mean values and the mean-value's stability of the covariance shaping adaptive filter are both better than those of the standard Kalman filter,and the positioning accuracy is increased by approximately 50％,and the sub-level single-point positioning navigation services can be provided.Compared with the centralized filter design scheme,the computational complexity of the distributed filter scheme is reduced by 63.5％.