|Table of Contents|

[1] Zhang Weiming, Zhang Yanhua, Zhong Shan, Du Gang, et al. New approach of Kalman filter to nonlinear system [J]. Journal of Southeast University (English Edition), 2005, 21 (1): 24-28. [doi:10.3969/j.issn.1003-7985.2005.01.006]
Copy

New approach of Kalman filter to nonlinear system()
非线性系统卡尔曼滤波新方法
Share:

Journal of Southeast University (English Edition)[ISSN:1003-7985/CN:32-1325/N]

Volumn:
21
Issue:
2005 1
Page:
24-28
Research Field:
Electronic Science and Engineering
Publishing date:
2005-03-30

Info

Title:
New approach of Kalman filter to nonlinear system
非线性系统卡尔曼滤波新方法
Author(s):
Zhang Weiming1 Zhang Yanhua1 Zhong Shan2 Du Gang1
1Department of Information Measurement Technology and Instruments, Shanghai Jiaotong University, Shanghai 200030, China
2The Second Academy, China Aerospace Science and Industry Corporation, Beijing 100854, China
张卫明1 张炎华1 钟山2 杜刚1
1上海交通大学信息检测技术与仪器系, 上海 200030; 2中国航天科工集团公司第二研究院, 北京 100854
Keywords:
nonlinear extended Kalman filter unscented Kalman filter unscented transformation
非线性 广义卡尔曼滤波 unscented卡尔曼滤波 unscented变换
PACS:
TN713
DOI:
10.3969/j.issn.1003-7985.2005.01.006
Abstract:
In order to achieve higher accuracy in nonlinear/non-Gaussian state estimation, this paper proposes a new unscented Kalman filter(UKF).It uses a deterministic sampling approach.We choose the unscented transformation(UT)scaling parameters α=0.85, β=2, l=0 to construct 2n+1 sigma points.These sigma points completely capture the mean and covariance of the Gaussian random variables of the nonlinear system Yi=F(Xi).Simulation results show that the posterior mean and covariance of the sigma points can achieve the accuracy of the third-order Taylor series expansion after having propagated through the true nonlinear system Yi=F(Xi).Extended Kalman filter(EKF)only can achieve the first-order accuracy.The computational complexity of UKF is the same level as that of EKF.UKF can yield better performance and higher accuracy than EKF.
为了在非线性、非高斯系统估计中获得更好的精度, 提出一种新的unscented卡尔曼滤波(UKF).采用确定性采样方法, 通过选择unscented变换中的参数α=0.85, β=2和l=0, 确定出2n+1个σ点, 使这些σ点完全符合非线性系统Yi=F(Xi)的高斯随机变量的均值和方差.仿真结果表明:σ点通过实际的非线性系统Yi=F(Xi)传递后, 其后验均值和协方差可以达到泰勒展开式的三阶精度, 广义卡尔曼滤波(EKF)只能达到一阶精度.该UKF滤波与EKF算法复杂度相近, 但具有比EKF更好的估计精度.

References:

[1] Sorenson H W.Kalman filtering:theory and applications [M].New York:IEEE Press, 1985.23-25.
[2] Julier S J, Uhlmann J K, Durrant-Whyte H F.A new approach for filtering nonlinear systems [A].In:Proceedings of the American Control Conference Seattle[C].Washington, 1995.1628-1632.
[3] Julier S J, Uhlmann J K.A new extension of the Kalman filter to nonlinear systems [A].In:Proc of AeroSense:the 11th Int Symp on Aerospace/Defence Sensing, Simulation and Controls[C].Orlando, Florida, 1997, 3068:182-193.
[4] Julier S J, Uhlmann J K.Reduced sigma point filters for the propagation of means and covariances through nonlinear transformations [A].In:Proceedings of the American Control Conference Anchorage [C].Alaska, 2002.887-892.
[5] Merwe R, Doucet A, Freitas N, et al.The unscented particle filter [R].Technical report CUED/F-INFENG/TR 380, Engineering Department of Cambridge University, 2000.
[6] Handschin J E, Mayne D Q.Monte Carlo simulation techniques to estimation the conditional expectation in multi-stage non-linear filtering [J].Estimation Int J Control, 1969, 9(5):547-559.
[7] Kadirkamanathan V, Li P, Jaward M H, et al.A sequential Monte Carlo filtering approach to fault detection and isolation in nonlinear systems [A].In:Proceedings of the 39th IEEE Conference on Decision and Control [C].Sydney, 2000.4341-4346.
[8] Wan E A, van der Merwe R.The unscented Kalman filter for nonlinear estimation [A].In:IEEE Proceedings of Symposium 2000 on Adaptive Systems for Signal Processing, Communication and Control[C].Lake Louise, Alberta, Canada, 2000.153-158.
[9] van der Merwe R, Wan E A.The square-root unscented Kalman filter for state and parameter-estimation [A].In:Proceedings of the International Conference on Acoustics, Speech, and Signal Processing[C].Salt Lake City, Utah, 2001.3461-3464.
[10] Haykin Simon.The unscented Kalman filter, Kalman filtering and neural networks [M].Wiley Publishing, 2001.221-280.

Memo

Memo:
Biographies: Zhang Weiming(1969—), male, graduate;Zhang Yanhua(corresponding author), male, professor, zhangyh@sjtu.edu.cn.
Last Update: 2005-03-20