Marginalized cubature Kalman filtering algorithm based on linear/nonlinear mixed-Gaussian model

作     者：Hu Yumei Hu Zhentao Jin Yong 胡玉梅;Hu Zhentao;Jin Yong

作者机构：School of AutomationNorthwestern Polytechnical University College of Computer and Information EngineeringHenan University 

出 版 物：《High Technology Letters》 (高技术通讯（英文版）)

年 卷 期：2018年第24卷第4期

页      面：362-368页

核心收录：

学科分类：07[理学] 08[工学] 

基　　金：Supported by the National Natural Science Foundation of China(No.61771006) the Open Foundation of Key Laboratory of Spectral Imaging Technology of the Chinese Academy of Sciences(No.LSIT201711D) the Outstanding Young Cultivation Foundation of Henan University(No.0000A40366) the Excellent Chinese and Foreign Youth Exchange Programme of China Science and Technology Association(2017CASTQNJL046) 

主　　题：state estimation marginalized modeling mixed-Gaussian model cubature Kalman filter 

摘      要：Aiming at improving the estimation accuracy and real-time of nonlinear system with linear Gaussian sub-structure,a novel marginalized cubature Kalman filter is proposed in Bayesian estimation framework. Firstly,the marginalized technique is adopted to model the target system dynamics with nonlinear state and linear state separately,and the two parts are estimated by cubature Kalman filter and standard Kalman filter respectively. Therefore,the linear part avoids the generation and propagation process of cubature points. Accordingly,the computational complexity is ***,the accuracy of state estimation is improved by taking the difference of nonlinear state estimation as the measurement of linear state. Furthermore,the computational complexity of marginalized cubature Kalman filter is discussed by calculating the number of floating-point operation. Finally,simulation experiments and analysis show that the proposed algorithm can improve the performance of filtering precision and real-time effectively in target tracking system.