DocumentCode :
1795245
Title :
On a new higher degree Cubature Kalman filter
Author :
Xiangyuan Huang ; Xiaqing Tang ; Meng Wu
Author_Institution :
Dept. of Control Eng., Acad. of Armored Force Eng., Beijing, China
fYear :
2014
fDate :
8-10 Aug. 2014
Firstpage :
2076
Lastpage :
2081
Abstract :
A new nonlinear Guassian filter named new 5th-degree cubature Kalman filter is proposed based on the conventional 3rd-degree CKF, for a higher filter accuracy, stronger numeric stability, acceptable computation complexity. The heart of new CKF is 5th-degree spherical and radial rule that a new sampling approach for points and corresponding weights. The 5th-degree CKF may provide a suboptimal solution for high-dimensional nonlinear system with higher accuracy than 3rd-degree and lower calculation than 5th-degree Gauss-Hermite quadrature filter. There is a comparison of filter accuracy and numeric stability in nonlinear initial alignment of SINS between 3rd, 5th and new CKFs. The results demonstrate that the new CKF has higher estamination accuracy and stronger stability with important application value.
Keywords :
Gaussian processes; Kalman filters; computational complexity; nonlinear filters; nonlinear systems; numerical stability; 3rd-degree CKF; 5th-degree Gauss-Hermite quadrature filter; 5th-degree cubature Kalman filter; SINS; computation complexity; estamination accuracy; filter accuracy; high-dimensional nonlinear system; nonlinear Guassian filter; nonlinear initial alignment; numeric stability; radial rule; spherical rule; suboptimal solution; Accuracy; Approximation methods; Estimation; Filtering algorithms; Kalman filters; Polynomials; Vectors;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Guidance, Navigation and Control Conference (CGNCC), 2014 IEEE Chinese
Conference_Location :
Yantai
Print_ISBN :
978-1-4799-4700-3
Type :
conf
DOI :
10.1109/CGNCC.2014.7007497
Filename :
7007497
Link To Document :
بازگشت