摘要
针对水下无人航行器(UUV)的航位推算导航方法(DR)和水下应答器(UTP)组合导航系统中传统滤波器因观测噪声统计模型不准确或未知而出现的滤波器发散问题,提出了一种基于变分贝叶斯的平方根容积卡尔曼滤波算法,该算法利用变分贝叶斯方法对DR/UTP组合导航系统的状态和时变观测噪声进行估计,并引入自适应调节因子来提高对观测噪声的逼近精度,然后利用平方根容积卡尔曼滤波对系统状态进行更新。仿真结果表明,该滤波算法能够较好地跟踪UUV的DR/UTP组合导航系统外部观测噪声方差的不断变化,可有效提高对DR/UTP组合导航系统各参数的估计精度。
Because of the inaccurate and time-variant measurement noise in statistical model, the conventional filters diverge severely in the dead reckoning（DR） navigation combined with the underwater transponder（UTP） for Unmaned Underwater Vehicle （UUV）. A square root cubature Kalman filtering algorithm based on variational Bayes is proposed to solve this problem. The proposed algorithm estimates the system states and time-variant measurement noise by variational Bayes, introduces an adaptive regulator to improve the approximation accuracy of measurement noise, and updates the system states by square root cubature Kalman fitering. The simulation results show that the algorithm can track the external measurement noise variance and improve estimation accuracy of parameters of the DR/UTP integrated navigation system effectively.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2016年第12期2743-2749,共7页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(51409055)
黑龙江省自然科学基金(E2015050)
黑龙江省博士后科研启动基金(3236310290)项目资助
关键词
DR/UTP组合导航
平方根容积卡尔曼滤波
变分贝叶斯
自适应
dead reckoning/underwater transponder positioning （DR/UTP） integrated navigation
square root cubature Kalman filtering
variational Bayes
adaptive
