Handling Particle Filter Sample Impoverishment for Orbit Determination
The purpose of this study is to present approaches for dealing with sample impoverishment, which is a common implementation difficulty when using particle filters. Solving this challenge can improve particle filter performance greatly, and can be the difference between success and failure. The number of actually unique sample values decreases, resulting in sample impoverishment. The difficulty is exacerbated when all of the particles eventually collapse to the same value. when there are mistakes in the modeling Increasing the number of particles is a straightforward option that can soon lead to unreasonably high processing demands and typically merely postpones the inevitable sample depletion. There are smarter approaches to deal with this problem, such as roughening and before editing, which will be covered in this article. The bootstrap filter is used to build recursive Bayesian filters in the nonlinear particle filter. The application involves using real data from GPS receivers to calculate the orbit of an artificial satellite. The orbital nonlinear problem Estimating values that completely specify the body trajectory in space, processing a set of data, such as space GPS receivers on-board the satellite, is the essence of determination. Nonlinear data (pseudo-ranges) can be extracted from this collection and used to determine the orbital state. The nonlinear particle filter is developed for the usual differential equations representing orbital motion and the GPS measurements equations, allowing the bootstrap to be used. The orbital state is also estimated using the technique. The evaluation will be done by comparing the bootstrap method outcomes obtained for each methodology that deals with sample impoverishment, as well as convergence speed and computational implementation complexity. The advantages and disadvantages of the implementations will be highlighted based on the study of such criteria.
Author (s) Details
Paula Cristiane Pinto Mesquita Pardal
USP (University of São Paulo) – EEL/LOB, Estrada Municipal do Campinho, s/n. CEP: 12602-810. Lorena, SP, Brasil.
Helio Koiti Kuga
ITA/DCTA (Technological Institute of Aeronautics), Praça Marechal Eduardo Gomes, 50. CEP: 12228‑900. São José dos Campos, SP, Brasil.
Rodolpho Vilhena de Moraes
UNIFESP (São Paulo Federal University) – ICT, Rua Talim, 330. Vila Nair – CEP: 12231-280. São José dos Campos, SP, Brasil.