Exploring an Effective Nonlinear Model Predictive Unscented Particle Filtering (NMPUPF) Algorithm
Yi Gao,Shesheng Gao,Zhihua Feng
DOI: https://doi.org/10.3969/j.issn.1000-2758.2012.05.019
2012-01-01
Xibei Gongye Daxue Xuebao/Journal of Northwestern Polytechnical University
Abstract:In navigation calculations, model error is assumed as some kind of noise for Gaussian white noise to process, thus causing large state estimation error of nonlinear filtering system and even divergence. We present a new NMPUPF algorithm based on predictive filtering and unscented particle filtering to improve the accuracy of navigation calculations. Section 1 of the full paper explains our NMPUPF algorithm, which we believe is new and effective. Section 1 gives a fairly detailed 6-step procedure for using our new algorithm. The core of the 6-step procedure consists of: (1) initialization, (2) calculating state estimation which includes the model error, (3) calculating estimation of systematical output vector, (4) calculating model error, (5) correcting nonlinear system model, (6) recycle calculations. Simulation results, presented in Figs. 1 through 4, and their analysis demonstrate preliminarily that the proposed algorithm is indeed effective for filtering calculations based on nonlinear and non-Gaussian system model, thus improving significantly the filter performance and integrated navigation calculation accuracy.