The singularity of the planar 3-RRR flexible parallel manipulator is complicated due to the closed-chain coupling effect. The parallel manipulator is easy to occur self-excited vibration near the inverse Jacobian singular configurations， which seriously affects the accuracy of the mechanism and damages its structure. In order to restore the parallel mechanism to normal operation， active vibration controllers are designed to suppress the self-excited vibration. Firstly， the inverse kinematics model of the mechanism is established， and the singularity judgement conditions of the parallel mechanism are obtained based on the velocity Jacobian matrix. On the basis of analyzing the mechanism of self-excited vibration， the vibration acceleration signal is filtered and phase shifted. Combining acceleration feedback and position error compensation， a fuzzy neural network nonlinear controller and BP neural network active disturbance rejection controller are designed. Finally， the effectiveness of the two intelligent controllers is verified by conducting active vibration control experiments. The experimental results show that the two control algorithms can quickly and effectively suppress the self-excited vibration while ensuring the positioning accuracy of the parallel manipulator.