Adaptive control design of neural fuzzy system for NARMA-L2 model

An adaptive neural control method is presented for the nonlinear discrete-time systems with the NARMA-L2 model. The neural fuzzy system is integrated with the approximate model-based control method to handle the nonlinear complexity, where the multiple fuzzy CMAC (MFCMAC) network is used to compensate the approximate NARMA model of the nonaffine nonlinear system. The weights of neural networks are modified by a novel adaptive algorithm, which guarantee the stability of the neural system without the persistent excitation requirement. The stability of the closed-loop system is proved with the Lyapunov method. Simulation results show that the method is effective.