Aiming at the problems of traditional filling quantitative control methods of liquid filling machine, such as low control stability and poor control accuracy, this paper introduces RBF neural network to high-precision filling quantitative control of liquid filling machines. The servo drive metering cylinder transfer function of liquid filling machine was constructed, and the high precision filling disturbance characteristic analysis of liquid filling machine was realized by combining the spatial perturbation fusion method. The parameter adaptive identification method is used for the quantitative analysis of the high-precision filling of the liquid filling machine, and the filling of the liquid filling machine is controlled by the B-spline curve fitting method. Through adaptive parameter adjustment, the RBF neural network model is constructed to realize the optimized design of high-precision filling quantitative control of the liquid filling machine. The simulation results show that the stability of liquid filling quantitative control of the proposed method is better, and the filling quantitative control of the filling machine is better. This method improves the ability of high-precision filling quantitative control of the liquid filling machine.