针对激光陀螺捷联惯导系统在动态尤其是高动态环境下的姿态误差显著增大的问题，提出了一种基于改进高斯混合粒子滤波的纯方位跟踪算法。算法基于混合粒子的卡尔曼滤波和粒子滤波的特点，用有限的高斯模型来近似后验状态密度、系统噪声和观测噪声的分布通过EM的算法设计实现模型的降阶，一定程度上克服了EM算法迭代的结果需要依赖初始值、可能收敛到局部最大点或可能收敛到参数空间边界的缺点，从而改善了粒子携带信息的衰减问题。通过仿真与试验结合，在纯动态应用环境下的姿态与定位精度补偿效果，与传统Kalman滤波相比，算法在保持高精度估计能力的同时，具有较强的鲁棒性，是解决非线性系统状态估计问题的一种有效方法。%A improved Gaussian mixture particle filter algorithm was proposed for the highly non -linear passive location and tracking system .The traditional Kalman filter system can not attain widely be applied in that the lack of model precision ,and can not realize the faulty check .The united filter was researched , the simulation research progress of the multi-model adaptive filters .From the simulation result analysis we know that multiple adaptive filter system can achieve stronger adaptability and higher accuracy than one of the single model in practice .The algorithm which based on the characteristics of SPKF and particle filter ,used limited Gaussian mixture model to approximate the posterior density of states ,system noise and measurement noise .Then the algorithm which used the greedy EM to obtain the reduction of model order , overcooked the disadvantage of the standard EM that assumed the number of the mixture components is a known a priori ,the performance of the overall parameter estimation process depends on the given good ini-tial settings,and the estimated parameter can be resulted from some local points .The effects caused by sampling depletion were lessened .SIN results show that the algorithm outperforms the one based on MPF , and stability .Therefore it is more suitable to the nonlinear state estimation .
展开▼
