Covariance intersection‐based fusion algorithm for asynchronous multirate multisensor system with cross‐correlation

Abstract

This study is concerned with asynchronous data fusion problem in the non-linear multisensor multirate system, where the observation noises are coupled with the system noise of the previous step. An estimation algorithm is designed based on covariance intersection (CI) fusion algorithm. First, the system is reformulated by using multiscale system theory. The local filtering algorithm processes the cross-correlated noises by exploiting the conditional Gaussian distributions. Furthermore, the first-order Stirling's interpolation algorithm and third-degree spherical-radial rule are used for non-linear approximate estimation. The fusion of the local estimate results is based on CI fusion algorithm. This study extends the currently available asynchronous data fusion algorithm for a multirate multisensor non-linear system, and considers the asynchronous correlation, then a novel algorithm is proposed, which has the ability to deal with the correlation, and is robust to the correlation between any two local filters. The effectiveness and superiority of the proposed algorithm are indicated by the results of numerical simulation.