基于凸优化技术的多目标鲁棒滤波组合导航方法

Kalman滤波在组合导航中的应用已很广泛,但是,在系统不确定情况下,误差出现发散,精度下降,采用的基于Riccati方程的H∞滤波技术对系统噪声的不确定性则具有一定的鲁棒性,但性能参数选取仅能凭经验而无系统的方法,给工程实践带来一定困难.根据现代控制理论最优控制理论思想,使用凸优化技术设计鲁棒混合H2/H∞滤波器融合高精度的GPS和CNS量测信息,解决系统不确定性问题,通过在惯性/卫星/天文组合系统中的半物理数据仿真进一步对混合H2/H∞滤波和H∞滤波算法进行了比较.仿真结果表明,真实器件噪声情况下,鲁棒混合H2/H∞滤波算法明显具有良好的精度.

Multiobjective Robust filtering Algorithm for Integrated Navigation System Based on convex optimization

Kalman filter is widely applied in integrated navigation system. When there exists uncertainty in system parameters or the statistics of the noises, the performance of the standard Kalman filtering will be greatly degraded, the precision drops. H∞robust filter based on the Riccati equation has robust ability, but its performance parameter is selected by experience and has no systematic method. It is difficult to carry out in practice. According to the optimal control theory and using the convex optimization techniques, an mix H2/H∞ robust filter based on LMI(Linear matrix inequality) with high accuracy GPS and the CNS measure information is presented. It can resolve model uncertainty and the noise non-Gauss question. The hardware in-the-loop simulation of SINS/CNS/GPS Integrated Navigation System is used to compare the precision. The simulation results indicate that muhiobject robust filter have better robustness and more precise than H∞ falter using real apparatus data.