摘要
基于采样方法的滤波器在非线性滤波领域内得到了广泛应用。其共同特点是利用抽样粒子点模拟系统状态的概率分布,从而不受状态先验分布假设(如高斯假设)的约束,拥有更高的滤波精度和更广的应用范围。在确定性抽样意义下论述了几种现有的确定性采样滤波器,并对这几种滤波器在状态估计领域(滤波)的应用进行了精度和计算负荷分析。采用一维及多维算例验证了几种方法的估计精度。给出基于实际系统需求一般性评价和选用原则。
Sampling-based filtering approaches are widely used for the on-line estimation problem of non-Gaussian nonlinear systems. They commonly take advantages of matching the prior density of system state by certain sampling strategies. Theoretical analysis and experimental results on several popular deterministic sampling filters were proposed to show performance on coping with non-gauss nonlinear filtering problems; moreover, guidelines and suggestions were given for practical appliance requisitions.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2007年第18期4265-4269,共5页
Journal of System Simulation
基金
国防科技"十五"预研(102010302)
国家自然科学基金(60404011)
校英才计划
校青年基金。
关键词
非线性滤波器
UKF
高斯厄米特滤波器
中心差分滤波器
分离差分滤波器
估计
确定性采样
nonlinear filter
Unscented Kalman Filter
Gaussian-Hermite Filter
Central Difference Filter
Divided Difference Filter
estimation
deterministic sampling