Abstract: This paper studies robust adaptive tracking for affine nonlinear system based on dynamic neural networks. The learning laws with respect to modeling error for unknown affine systems are proposed using the Lyapunov synthesis approach with the projection modification method which does not require a priori knowledge of norm for ideal weight matrices. The δ projrction and hysteresis technology are used, which is allowed to guarantee the stability of the resulting controller. Simulation results are given to verify the effectiveness of the newly proposed DNN adaptive contral algorithm.