蒙特卡罗粒子滤波算法应用研究

Study the Application of Monte Carlo Particle Filter Algorithms

随着这些年计算机硬件水平的发展,计算速度的提高,源自序列蒙特卡罗方法的蒙特卡罗粒子滤波方法的应用研究又重新活跃起来。本文的这种蒙特卡罗粒子滤波算法是利用序列重要性采样的概念,用一系列离散的带权重随机样本近似相应的概率密度函数。由于粒子滤波方法没有像广义卡尔曼滤波方法那样对非线性系统做线性化的近似,所以在非线性状态估计方面比广义卡尔曼滤波更有优势。在很多方面的应用已经逐渐有替代广义卡尔曼滤波的趋势。

As the computer hardware developing and the fast advances of computers in the last several years, Monte Carlo particle filter algorithms, which origin from Monte Carlo methods, have become popular again. The Monte Carlo particle filter algorithms in this paper use the concepts of sequential importance sampling. The base idea of particle filter is the approximation of relevant probability distributions using a set of discrete random samples with associated weights. The particle filter is a powerful nonlinear estimation method and has been shown to be a superior alternative to the EKF in a variety of applications, because Particle Filter does not involve the linearization approximating of nonlinear systems, that is required by the EKF.