Self-adaptive volume Kalman filtering method

Abstract

The invention relates to a self-adaptive volume Kalman filtering method, and in particular relates to a self-adaptive volume Kalman filtering method with a fading memory time-change noise statistic estimator. The method comprises the following steps of: (1) setting initial parameters; (2) updating the time; (3) updating the measurement; (4) constructing the fading memory time-change noise statistics estimator; and (5) estimating and modifying the noise in real time. Compared with a standard volume Kalman filtering method, the method does not demand to make the prior statistic characteristics of the known noise precise and has the self-adaptive capability to copy with the noise change; and moreover, the recursion formula of the noise statistic estimator is simple and easy to realize, and is unbiased to noise statistic estimation.

Description

technical field [0001] The invention relates to an adaptive volume Kalman filtering method, in particular to an adaptive volume Kalman filtering method with a time-varying noise statistical estimator with fading memory. Background technique [0002] Canadian scholar Arasaratnam proposed a new nonlinear filtering method in the document "Cubature Kalman Filters" (IEEE Transactions on Automatic Control, 2009, 54(6): 1254-1269): volumetric Kalman filter (Cubature Kalman Filter, namely CKF). Volumetric Kalman filtering is a filtering algorithm derived from strict mathematical derivation based on Bayesian theory and volumetric rules. It is theoretically guaranteed. According to the volumetric criterion, a group of points with the same weight are transformed through nonlinear system equations. Generate new points to predict the state of the system at the next moment, avoiding the linearization of the nonlinear model, and its accuracy reaches the third order. Due to the advantages...

AI Technical Summary

Problems solved by technology

However, in practical applications, the prior statistics of noise are unknown or inaccurate. Even if the prior statistical characteristics of noise are known, due to the influence of internal and external uncertain factors, the statistical characteristics of noise are easily changed, showing relatively Strong time-varying properties

The standard volumetric Kalman filter does not have the adaptive ability to deal with noise statistical changes, which may lead to large estimation errors and even filter divergence when the noise statistics are unknown and time-varying.

Images