Data-driven Kalman filter

Details

Supervisors

Bianchin, Gianluca

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en mathématiques appliquées, à finalité spécialisée

Abstract

This thesis investigates a newly proposed data-oriented variant of the Kalman Filter. Unlike traditional model-based approaches, this method relies directly on input-output data, avoiding the need for an explicit model of the system. Such data-driven filtering techniques open promising avenues in scenarios where system identification is challenging due to internal complexity or limited model knowledge. We focus on the application of the new variant on linear and nonlinear systems to highlight its performances compared to classical model-based approaches.