摘要
提出一种新的渐消自适应Unscented粒子滤波算法,通过Sigma点来获取状态估值和协方差阵,利用渐消因子自适应的调节权值大小,得到一种参数可调节的重要性密度函数。该重要性密度函数考虑了最新量测的影响,更合理地利用有效信息,保证了粒子多样性,使滤波性能明显改善,能更好地解决非线性非高斯系统模型的滤波问题。将提出的算法应用于SINS/SAR组合导航系统中,与扩展Kalman滤波和渐消自适应扩展Kalman滤波比较,仿真结果表明,提出的滤波算法能提高导航解算的精度,其性能明显优于扩展Kalman滤波和渐消自适应扩展Kalman滤波。
We present a fading adaptive UPF algorithm by adopting the concept of unscented transform and the fading factor in particle filtering. This algorithm makes dissemination of information more reasonable and overcomes the limitations of the general particle filtering by using sigma point to obtain state estimation and covariance, and then the fading factor can adaptively regulate the weight function. Thus it provides a reliable importance density function and is suitable for filtering calculations based on nonlinear and non-Gaussian models, through considering the latest measurement information and ensuring the particle diversity. The proposed algorithm is applied to SINS/ SAR integrated navigation system. Simulation results, presented in Figs. 1,2 and 3, and their analysis demonstrate preliminarily that the fading adaptive UPF algorithm outperforms the extended Kalman filtering and fading adaptive extended Kalman filtering ones in terms of accuracy, thus improving the precision in navigation system.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
2012年第1期27-31,共5页
Journal of Northwestern Polytechnical University
基金
航空科学基金(20080818004)
陕西省自然科学基金(SJ08F04)资助
关键词
Unscented粒子滤波
渐消滤波
渐消自适应Unscented粒子滤波
组合导航
adaptive filter, algorithms, analysis, calculations, computer software, errors, functions, improvement, inertial navigation systems, Kalman filtering, measurements, models, Monte Carlo methods, nonlinear systems, reliability, sampling, simulation, state estimation, synthetic aperture radar, velocity
fading adaptive UPF( unscented particle filtering), integrated navigation