An adaptive neural network control scheme is proposed for a class of interconnected systems with state and input unmodeled dynamics as well as input nonlinearity. The unknown nonlinear continuous functions are approximated by radial basis function neural networks(RBFNNs). State and input unmodeled dynamics are dealt with by introducing dynamic signal and normalization signal. The strict-feedback system with time-varying output constraint is transformed into a novel strict-feedback system without constraint by introducing one to one nonlinear mapping. By theoretical analysis, all the signals in the closed-loop control system are proved to be semi-globally uniformly ultimately bounded. Simulation results show the effectiveness of the proposed approach.