Adaptive identification and prediction control for time delay nonlinear systems based on neural networks

This paper presents the identification, prediction and control design for nonlinear strict-feedback systems with an input time-delay. The system is firstly transformed into a normal form by defining new state variables. A dynamical identification with a neural network (NN) is proposed to estimate the system states. The predictive NN weights are obtained without iterative calculations and utilized in constructing the adaptive predictor. Feedback control design using the predictive states is finally studied. Simulations are included to validate the effectiveness of the proposed method.