In this paper, a recurrent perturbation fuzzy neural network (RPFNN) is used to online approximate an unknown nonlinear term in the system dynamics. A sine-cosine perturbed membership function is used to handle rule uncertainties when it is hard to exactly determine the grade of the value of fuzzy sets. Unlike type-2 fuzzy sets use an extra type reduction operation to find the output, the proposed RPFNN does not require heavy computational loading. Meanwhile, this paper proposes an intelligent dynamic sliding-mode neural control (IDSNC) system which is composed of a neural controller and an exponential compensator.