离散系统滤波非线性近似方法综述

Overview of Nonlinear Approximation Techniques for Discrete System Filtering

从离散系统的最优贝叶斯估计原理出发,根据对后验密度近似计算方法不同,将近似方法分为3类:函数近似方法、采样近似法和高斯和近似法。详述了这些近似方法的原理、特点及采用这些近似方法进行状态估计的非线性滤波的特点及缺陷。最后指出了非线性滤波的发展方向。

Based on study to the principle of discrete system s optimal Bayesian estimation,and according to different approximation techniques of posterior density,the authors categorize these approximation approaches into three types： function approximation,sampling-based approximation,and Gaussian sum approximation.The principles and characteristics of the above approximation approaches are reviewed in detail,the advantage and disadvantage of nonlinear filtering that use these approximation techniques for state estimation are described. Further prospects of nonlinear filtering are introduced.