The piezoelectric ceramic injection valve is the core executive component of the dispensing robot, and the tightness of its striker and nozzle will affect the dispensing frequency, the volume of glue point, and the amount of glue in a single point and so on. In the prior art, the tightness of the top tightening of the striker and nozzle is manually adjusted according to the operation experience. This method is time-consuming and cannot keep the same tightness every time. In order to overcome the shortcomings of the existing technology, this paper designs a method to adjust the tightness of the top tightening of the striker and nozzle of piezoelectric ceramic injection valve based on the current sensor. By collecting the load current of the controller in real time, the model between the current value and the corresponding screw sleeve rotation angle is established offline by using the improved BP neural network. The relative value of the tightness is obtained via the transformation of the angle value, so that the tightness of each adjustment is consistent to ensure the same displacement of piezoelectric ceramics. The experimental results show that the established model can basically ensure the consistency of tightness according to the relative value, and realize the visual adjustment.