A local learning approach to real-time parameter estimation - application to an aircraft

This paper proposes an approach based upon local learning techniques and real-time parameter estimation, to tune an aircraft sideslip estimator using radial-basis neural networks, during a flight test. After a presentation of the context, we recall the local model approach to radial-basis networks. The application to the estimation of the sideslip angle of an aircraft, is then described and the various results and analyses are detailled at the end before suggesting some improvement directions.