Approximated Particle Flow for Continuous Filter, Predictor, and Smoother

抄録

The filtering and smoothing techniques for nonlinear continuous-discrete model which is applicable to physical state estimation problems are explained. The multiple distribution estimation and distribution selection are applied to design the non-Gaussian filter. After the execution of the filter, the non-Gaussian smoother is applied to improve the filtered state estimates further. The filtering and smoothing problem for the spacecraft re-entry simulation model is provided as the numerical example and the proposed approach is compared with the other non-Gaussian filters and smoothers.