Data Fusion with Model Error Estimators and Stability Analysis

摘要

An approach for sensor data fusion that utilizes model error estimators is presented. These estimators are based on the method of invariant embedding and H-infinity concept; and both continuous time and discrete time nonlinear systems are evaluated. Also, an observer for nonlinear continuous time dynamic system is proposed that utilizes the gain and the associated matrix Riccati type differential equation from the combined invariant embedding (IE) and H-infinity (HI) theories. Then, Lyapunov energy (LE) functional is used for deriving the condition for the local asymptotic stability for the observer's error dynamics. The performances of the model error estimators-based data fusion scheme and of the continuous observer are evaluated by simulation carried out in MATLAB. The presented results validate the theoretical asymptotic behaviour of the nonlinear observer, as well as the convergence properties of the data fusion scheme, since both are based on IE/HI theory. This type of validation study is a novel feature of this contribution.