基于新型滤波器-HABF的SINS传递对准仿真

Simulation of SINS transfer alignment based on hybrid annealed particle filter

传统动基座传递对准主要采用扩展卡尔曼滤波技术。但在动基座传递对准的非线性、非高斯条件下,这种基于模型线性化和高斯假设的滤波方法在估计系统状态及其方差时误差较大且可能发散。混合退火粒子滤波针对非线性、非高斯系统状态的在线估计问题,提出一种新的基于序贯重要性抽样的粒子滤波算法。在滤波算法中,用状态参数分解和退火系数来产生重要性概率密度函数,此概率密度函数综合考虑了转移先验、似然、噪声的统计特性以及最新的观察数据,因此更接近于系统状态的后验概率。实验仿真结果表明,这种基于混合退火粒子滤波器不仅比扩展卡尔曼滤波提高了传递对准的精度,而且又比传统的粒子算法减少了时间。

Traditional method for moving-base transfer alignment generally applies the technology of Extended Kalman Filtering（EKF） which is based on the approximately-linear and Gaussian model. But under the nonlinear and non-Gaussian condition in moving-base transfer alignment, larger errors will be generated, even leading to divergence. So a new particle filter, Hybrid Annealed Particle Filter （HAPF）, was proposed which was based on the sequential importance sampling （SIS） for the on-line estimation of non-Gaussian nonlinear systems, In this filtering method, the state parameters separation and the annealing parameter are used to produce importance function. Since the distribution function makes full use of the prior, likelihood, statistical characteristics of noise and the newest observation data, it is much closer to posterior distributions, Simulations show that the proposed particle filter is higher than EKF in precision, and shorter than the conventional particle algorithm in time.