Identification and control of time-delay system with recurrent wavelet neural networks

A recurrent wavelet neural network (RWNN) is introduced to realize identification and control of system with time delay. The identification is based on model type of nonlinear auto-regressive with exogenous inputs (NARX). The method incorporates the delayed massage of the system, the resulting model can give predictions to the object system. The wavelet-network-based identification model is used for online system identification, and the experiment result of predictive ship course predictive control proved the efficiency of the recurrent neural network model. The identification results are implemented in a control strategy and the simulation result shows the effectiveness of the proposed identification and control method.