An enhanced anti-disturbance filtering scheme for non-Gaussian systems with additional exogenous disturbances

摘要

In this paper, we study the state estimation problem for a general class of nonlinear systems subject to both non-Gaussian noise and exogenous disturbance. The exogenous disturbance under consideration can be treated as the output of a linear exogenous system. For this problem, direct application of the traditional particle filter (PF) may require an excessive number of particles and hence lead to prohibitive computation cost. To strick a balance between estimation accuracy and computational efficiency, we borrow the idea of disturbance compensation and propose an enhanced anti-disturbance filtering scheme where state augmentation is avoided by deriving an analytical estimate of the conditionally linear state. In the proposed scheme, we only need to generate particles from a lower dimensional state space, and the computational burden is thereby released to a large extent. Simulation results demonstrate that, under equal computational burden, our method can achieve higher accuracy than the traditional PF method.