An Instrumental Variable Method for Nonlinear System Identification Based on Sliding Modes

International audience; The paper deals with the dynamic identification in noisy environment of a class of nonlinear systems whose parameters depend linearly on the unknown parameters. The identification algorithm is based on an extended sliding modes observer and a least square estimation using an instrumental variable technique. The study of the measurement noise effect on the measured signals shows that the proposed solution is consistent. An application on a permanent magnet synchronous machine highlights the efficiency of the technique.