电光与控制, 2009, 16 (12): 48, 网络出版: 2010-09-08
常用非线性滤波方法比较研究
A Comparison of Several Widely Used Nonlinear Filtering Approaches
非线性状态估计 扩展卡尔曼滤波 Unscented卡尔曼滤波 粒子滤波 nonlinear state estimation Extended Kalman Filter (EKF) Unscented Kalman Filter(UKF) Particle Filter(PF)
摘要
几乎所有的现实系统都是非线性的,因而非线性估计问题显得尤为重要。通过分析3种常用的非线性滤波器:扩展卡尔曼滤波器、Unscented卡尔曼滤波器以及粒子滤波器的原理,确定其适用性。用单变量非平稳模型及再入目标跟踪模型,通过Monte-Carlo仿真计算估计的均方误差及时耗,进而对上述3种滤波器滤波精度、一致性以及运算量进行比较。
Abstract
Estimation of nonlinear systems is extremely important because almost all practical systems involve nonlinearities of one kind or another. Three kinds of widely used methods for estimation of nonlinear dynamic system, i.e., Extended Kalman Filter (EKF), Unscented Kalman Filter (UKF), and Particle Filter (PF), are studied here. Particle Filter is a new filtering method based on Bayesian estimation and Monte-Carlo method and can effectively cope with complicated nonlinear and/or non-Gaussian problems. Univariate Nonstationary Growth Model (UNGM) and Reentry Vehicle Tracking (RVT) were used in comparison of the performance of the above filters. Through Monte-Carlo simulations, the average mean square error matrix and running time were calculated out to analyze the accuracy, statistical consistency and the computational speed of these estimators.
蔚国强, 杨建业, 张合新. 常用非线性滤波方法比较研究[J]. 电光与控制, 2009, 16(12): 48. YU Guoqiang, YANG Jianye, ZHANG Hexin. A Comparison of Several Widely Used Nonlinear Filtering Approaches[J]. Electronics Optics & Control, 2009, 16(12): 48.