In order to decrease the influence of sting vibration on wind tunnel testing data, an active vibration reduction system using piezoelectric actuators is developed. The variation and complicated nature of the sting system make it challenging to model and determine the control parameter for a conventional PID. To solve this problem, a new control method combining a neural network and conventional PID is proposed. The performance of the control system is then validated through both ground tests and wind tunnel tests using balance signal feedback. Results show that decay time can be shortened by 50% compared to using a conventional PID, and the first mode amplitude is damped at least 19 dB under various wind speeds and attack angles.