[en] In this paper, a neural network controller that can be implemented in parallel with a PD controller is suggested for motion control of a hydraulic servo cylinder. By applying a self-excited oscillation method, the system design parameters of open loop transfer function of servo cylinder system are identified. Based on system design parameters, the PD gains are determined for the desired closed loop characteristics. The neural network is incorporated with PD control in order to compensate the inherent nonlinearities of hydraulic servo system. As an application example, a motion control using PD-NN has been performed and proved its superior performance by comparing with that of a PD control