Constrained Dynamic Systems Estimation Based on Adaptive Particle Filter

摘要

For the state estimation problem, Bayesian approach provides the most general formulation. However, most existing Bayesian estimators for dynamic systems do not take constraints into account, or rely on specific approximations. Such approximations and ignorance of constraints may reduce the accuracy of estimation. In this paper, a new methodology for the states estimation of constrained systems with nonlinear model and non-Gaussian uncertainty which are commonly encountered in practice is proposed in the framework of particles filter. The main feature of this method is that constrained problems are handled well by a sample size test and two particles handling strategies. Simulation results show that the proposed method can outperform particles filter and other two existing algorithms in terms of accuracy and computational time.