Square root simplex sampling cubature Kalman filtering method

Abstract

The invention provides a square root simplex sampling cubature Kalman filtering method. The method comprises calculating a system prediction state and a covariance matrix successively; calculating prediction measuration and a covariance matrix; calculating filtering gains; performing updating to obtain the system state and the covariance matrix. By the aid of the method, the problem that covariance matrixes are not positive definite caused by round-off errors of computers is solved by introducing the square root filtering technology, the filtering method robustness is improved, and the calculation amount can be greatly reduced by the simplex sampling method, and the filtering accuracy is improved compared with traditional filtering methods.

Description

technical field [0001] The invention belongs to the field of nonlinear filtering, and in particular relates to a square root monomorphic volume Kalman filtering method. Background technique [0002] Nonlinear filtering is one of the research hotspots in modern control, and it is widely used in engineering applications such as target tracking, satellite orbit / attitude estimation, and detector navigation and control. Bayesian filtering is a method that can be solved recursively through appropriate assumptions and approximations. It is the mainstream idea to solve nonlinear filtering problems. It has also been widely studied and applied, and a series of results have been achieved. [0003] Among them, the Extended Kalman Filter (EKF) algorithm is a traditional algorithm of the Bayesian filtering method. It performs a first-order linearization approximation to the nonlinear function through the Taylor expansion, ignoring the nonlinear characteristics of the system. When the init...

AI Technical Summary

Problems solved by technology

However, there are various rounding errors in numerical calculations during the filtering recursion process, which cannot ensure numerical stability.

Images