Gaussian Sum High Order Unscented Kalman Filtering Algorithm
WANG Lei1, CHENG Xiang-hong2, LI Shuang-xi1
1. School of Electrical and Electronic Engineering, Anhui Science and Technology University, Bengbu, Anhui 233100, China;
2. School of Instrument Science and Engineering, Southeast University, Nanjing, Jiangsu 210096, China
为了提高非线性变换的近似精度，提出了一种高阶无迹变换（High order Unscented Transform，HUT）机制，利用HUT确定采样点并进行数值积分去近似状态的后验概率密度函数，建立了高阶无迹卡尔曼滤波（High-order Unscented Kalman Filter，HUKF）算法.进一步的为了解决非线性、非高斯系统的状态估计问题，将HUKF与高斯和滤波（Gaussian Sum Filter，GSF）相结合，提出了一种高斯和高阶无迹卡尔曼滤波算法（Gaussian Sum High order Unscented Kalman filter，GS-HUKF），该算法的核心思想是利用一组高斯分布的和去近似状态的后验概率密度，同时针对每一个高斯分布采用高阶无迹卡尔曼滤波算法进行估计.数值仿真实验结果表明，提出的HUT机制与普通的无迹变换（Unscented Transform，UT）相比，具有更高的近似精度；提出的GS-HUKF与传统的GSF以及高斯和粒子滤波器（Gaussian Sum Particle Filter，GS-PF）相比，兼容了二者的优点，即具有计算复杂度低和估计精度高的特性.
A novel high order unscented transform (HUT) mechanism is proposed to improve the approximation accuracy of the nonlinear transformation.The HUT is adopted to select the Sigma points which can be used to approximate the posterior probability density of state variable by numerical integration.Thus the high order unscented Kalman filtering (HUKF) algorithm can be made up.Further,to solve the state estimation problem for nonlinear/non-Gaussian system,Gaussian sum high order unscented Kalman filter (GS-HUKF) is proposed by combining the HUKF and Gaussian sum filter (GSF).The basic idea of the GS-HUKF is that a cluster of Gaussian distribution is used to approximate the posterior probability density of state variable.At the mean time,each separated Gaussian distribution is estimated by HUKF.Numerical simulation results demonstrate that the proposed HUT has higher estimation precision than ordinary unscented transform (UT) method.The GS-HUKF has integrated advantages with respect to estimation accuracy and computational complexity and its performance is superior to the existing Gaussian sum filters.
[1] Pei H L,Sanjeev A,Tharaka A L,Thushara D A.A Gaussian-Sum based cubature Kalman filter for bearings-only tracking[J].IEEE Transactions on Aerospace and Electronic Systems,2013,49(2):1161-1176.
[2] Lim,K P.Ranking market efficiency for stock markets:A nonlinear perspective[J].Physica A:Statistical Mechanics and its Applications,2007,376(10):445-454.
[3] Zhang H,Feng G,Han C. Linear estimation for random delay systems[J]. Systems and Control Letters,2011,60(7):450-459.
[4] Li W,Jia Y. Distributed consensus filtering for discrete-time nonlinear systems with non-Gaussian noise[J].Signal Processing,2012,92(10):2464-2470.
[5] Schei T S.A finite-difference method for linearization in nonlinear estimation algorithms[J].Automatica,1997,33(11):2053-2058.
[6] Julier S J,et al.A new approach for filtering nonlinear systems[A].Proceedings of the American Control Conference[C].Seattle:IEEE,1995. 1628-1632.
[7] Arasaratnam I,Haykin S,Elliott R J.Discrete-time nonlinear filtering algorithms using Gauss-Hermite quadrature[J].Proceedings of the IEEE,2007,95(5):953-977.
[8] Arasaratnam I,Haykin S.Cubature kalman filters[J].IEEE Transactions on Automatic Control,2009,54(6):1254-1269.
[9] Ito K,Xiong K.Gaussian filters for nonlinear filtering problems[J].IEEE Transactions on Automatic Control,2000,45(5):910-927.
[10] Brunke S,Campbell M.Square root sigma point filtering for real-time,nonlinear estimation[J].Journal of Guidance,Control,and Dynamics,2004,27(2):314-317.
[11] Li W,Jia Y.Distributed consensus filtering for discrete-time nonlinear systems with non-Gaussian noise[J].Signal Processing,2012,92(10):2464-2470.
[12] Alspach D L,Sorenson H W.Nonlinear Bayesian estimation using Gaussian sum approximations[J].IEEE Transactions on Automatic Control,1972,17(4):439-448.
[13] M Sanjeev Arulampalam,et al.A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking[J].IEEE Transactions on Signal Processing,2002,50(2):174-188.
[14] Julier S J,et al.Unscented filtering and nonlinear estimation[J].Proceedings of the IEEE,2004,92(3):401-422.
[15] Ponomareva K,Date P,Wang Z.A new unscented Kalman filter with higher order moment-matching[A].Proceedings of Mathematical Theory of Networks and Systems[C].Budapest:E tv s Loránd University (ELTE),2010.1609-1613.
[16] Tenne D,Singh T.The higher order unscented filter[A].Proceedings of the IEEE American Control Conference[C]. New York:IEEE,2003. 2441-2446.
[17] Arasaratnam I,Haykin S,Elliott R J.Discrete-time nonlinear filtering algorithms using Gauss-Hermite quadrature[J].Proceedings of the IEEE,2007,95(5):953-977.
[18] Caputi M,Moose R L.A modified Gaussian sum approach to estimation of non-Gaussian signals[J].IEEE Transactions on Aerospace and Electronic Systems,1993,29(2):446-451.
[19] Clark J M C,Kountouriotis P A,Vinter R B.A Gaussian mixture filter for range-only tracking[J].IEEE Transactions on Automatic Control,2011,56(3):602-613.
[20] Horwood J T,Poore A B.Adaptive Gaussian sum filters for space surveillance[J].IEEE Transactions on Automatic Control,2011,56(8):1777-1790.
[21] Wu Y,Hu D,Wu M,et al.A numerical-integration perspective on Gaussian filters[J].IEEE Transactions on Signal Processing,2006,54(8):2910-2921.
2017, Vol. 45 
