Nonlinear system identification and application based on local model networks

Abstract

Local model networks based identification algorithm is developed and its application is considered for nonlinear systems. The characteristics of nonlinear system are approximated by combination of local models, and each model is followed by an appropriate weight coefficient corresponding to the operating condition. The structure selection of local models, the partition of operating range and estimation algorithm for local model parameters are illustrated, and the effectiveness of the proposed approach is demonstrated through application examples.