摘要
提出了一种适用于线性和非线性系统的带多步随机延迟量测高斯滤波器的一般框架解.为了完成状态的递归更新估计,噪声向量和先前时刻状态向量被扩展到当前时刻状态向量中.然后基于贝叶斯方法推导了扩展后状态向量的一般框架解.对于非线性系统,通过利用不同的数值计算方法计算贝叶斯解中的高斯加权积分可以推导获得不同的高斯近似滤波器.最后本文利用三阶球径容积准则来实施提出的方法,并通过量测被随机延迟多步的目标跟踪模型对所提出的方法进行了仿真,仿真结果验证了提出方法的有效性和优点.
This paper provides a general framework solution to state estimation of Gaussian filter for both linear and nonlinear dynamic systems with multiple step randomly delayed measurements. Noise and previous state vectors are added into the current state vector to facilitate its recursive update estimation. A general framework of Bayesian solution to the augmented state estimation is then derived. For nonlinear systems, different Gaussian approximation filters can be developed by utilizing different numerical methods for computing Gaussian weighted integrals involved in the Bayesian solution. Finally, the third-degree spherical-radial cubature rule is used to implement the proposed method. Simulation is performed based on a target tracking model, in which measurements are randomly delayed for multiple steps. The simulation results illustrate the efficiency and advantages of the proposed method.
出处
《自动化学报》
EI
CSCD
北大核心
2015年第1期122-135,共14页
Acta Automatica Sinica
基金
国家自然科学基金(61001154
61201409
61371173)
中国博士后科学基金(2013M530147)
黑龙江省博士后基金(LBH-Z13052)
哈尔滨工程大学中央高校基本科研业务费专项基金(HEUCFX41307)资助~~
关键词
一般框架
高斯滤波器
多步随机延迟量测
贝叶斯估计
General framework
Gaussian filter
multiple step randomly delayed measurements
Bayesian estimation