基于CKF的GNSS/INS列车组合定位鲁棒滤波算法

针对列车组合定位融合估计的非线性以及鲁棒性问题, 分析了GNSS/INS列车组合定位的基本原理和系统结构, 给出了非线性滤波算法CKF的基本过程, 通过将H_∞鲁棒滤波思想应用于标准CKF, 提出了一种基于CKF的新型鲁棒滤波算法, 从滤波器收敛判别及误差H∞范数界鲁棒性判别两方面探讨了滤波器的自主状态监测, 并采用青藏铁路实测数据对算法进行了验证。分析结果表明: 在该实测数据条件下, 所提算法平均估计误差与标准UKF、CKF相比, 分别降低了7.13%、4.85%;约束水平从31.5到1315.0时, 估计误差均方根增大了8.56%, 估计精度变化趋势较平缓, 该算法有效。

CKF-based robust filtering algorithm for GNSS/INS integrated train positioning

Based on the fusion estimation problems of nonlinearity and robustness in integrated train positioning, the principle and architecture of global navigation satellite system/inertial navigation system(GNSS/INS)integrated train positioning system were introduced, the scheme of CKF nonlinear filter approach was given, and a novel CKF-based robust filtering algorithm was presented by applying H∞ robust filter theory to standard CKF frame. The autonomous filter state monitoring was discussed for the proposed algorithm. The filter convergence criterion and the robustness criterion of error's H∞ norm bound were analyzed, and field tests in Qinghai-Tibet Railway were illustrated. Analysis result shows that under field test condition, the average estimation error of the proposed algorithm decreases by 7.13% and 4.85% respectively in comparison with standard UKF and CKF. The root mean square of estimation error increases by 8.56% with constraint level from 31.5 to 1 315.0, so the estimation precision changes mildly, and the algorithm is effective.