捷联惯导系统初始对准中IUPF的应用与设计

Application and design of iterative unscented particle filter for initial alignment of SINS

针对大失准角条件下捷联惯导系统误差模型的非线性引起的初始对准误差问题,引入了粒子滤波技术.首先,利用加性四元数误差建立了大失准角条件下的系统误差模型.然后,将无迹变换算法和迭代算法引入粒子滤波方法中,提出了迭代无迹粒子滤波算法.并且,为了解决迭代无迹粒子滤波算法中由于粒子数量所导致的算法精度和算法实时性的矛盾,采用采样重要性重采样的方法对迭代无迹粒子滤波算法进行修正,提出了一种既具有迭代无迹粒子滤波精度又计算量较小的新的非线性滤波算法.最后,进行了半物理仿真和数字仿真,比较了经典粒子滤波、无迹粒子滤波和修正后的迭代无迹粒子滤波等多种方案的滤波效果.仿真结果表明,在大失准角条件下,采用修正后的迭代无迹粒子滤波方法可以有效提高初始对准精度和算法的实时性.

The particle filtering is introduced to deal with the initial alignment error,which is caused by the non-linear model of strapdown inertial navigation system（SINS） with large misalignment angle.Firstly,on the basis of additional quaternion error,the error model of SINS is built up.Secondly,the iterative unscented particle filter（IUPF） based on unscented transformation（UT）,iterative filtering and particle filtering is constructed.In order to solve the contradiction of precision and real-time caused by the particle quantity of IUPF,the idea of sampling importance resampling（SIR） is applied to improve IUPF.The corrected IUPF has not only the same accuracy as the IUPF but also lower computational cost than the IUPF.Finally,some schemes of initial alignment by respectively using a partical filter,a unscented particle filter,a corrected IUPF and so on are compared.The simulation results indicate that the precision and real-time of initial alignment can be improved by the corrected IUPF under large misalignment angle.