Target tracking method based on Markov chain Monte-Carlo particle filtering

Abstract

The invention provides a target tracking method based on Markov chains Monte-Carlo particle filtering, which comprises steps of: 1, obtaining a group of initial particles from initial distribution and setting the initial mean value and variance of the initial particles at initial time; 2, sampling importance; 3, updating a weight number; 4, obtaining a normalized weight number; 5, resampling; 6, introducing an MCMC (Markov Chains Monte-Carlo) movement step; and 7, updating status. By the MCMC movement step, the invention pushes particles to an area with larger prior distribution and posterior distribution, improves the diversity of the particles and inhibits the depletion problem of a sample to some extent. The solvent of the depletion problem of the sample ensures the effect of algorithm resample so as to further enhance the precision of filtering. The MCMC movement step is easy to realize, thereby being capable of combining with other improvement steps to optimize the particle filtering. The MCMC movement step is added to increase the workload of a filtering method and decreases number of particles needed in accurate estimation, thereby enhancing the filtering efficiency.

Description

technical field [0001] The invention provides a function tracking method, and in particular relates to a particle filter method (PF-MCMC) which combines Markov chain Monte Carlo moving steps. Background technique [0002] Nonlinear filtering methods are widely used in signal processing, navigation guidance, target tracking, financial analysis, artificial intelligence and other related fields. The earliest nonlinear filtering algorithm is the Extended Kalman Filter (EKF). The core idea of ​​EKF is to make a linear approximation to the nonlinear model of the stochastic system, and its noise is based on the Gaussian assumption. Therefore, for the strong nonlinear non-Gaussian model The filtering effect is not good. In engineering applications, EKF is only effective for some specific models, and it cannot guarantee convergence and filtering accuracy for many nonlinear systems. Unscented Kalman filtering (UKF) is also a commonly used nonlinear filtering algorithm. UKF does not n...

AI Technical Summary

Problems solved by technology

[0003] The most common problem with PF is the phenomenon of particle degeneracy, that is, after several iterations, all particles except one have only tiny weights, which means that a lot of computational work is used to update those values ​​of the posterior probability density estimated to have little effect on the particles

A key technology to solve this problem is the re-sampling strategy. The basic idea is to re-sample N times through the posterior probability density function to generate a new particle set. Since the re-sampling is independent and identically distributed, the weight of the particles is reset to 1 / N, the negative problem brought by resampling is the phenomenon of sample depletion, that is, particles with high weights are copied many times, and the sampling results contain many repeated points, which cannot effectively reflect the probability distribution of state variables, thus losing the particle's Diversity, which can even lead to filter divergence

Images