In order to ensure that the various functions of the pillow type packing machine can be run accurately and efficiently, the control strategy of neural network is researched particularly. Through modeling and simulation, the advantages and disadvantages of classical Proportion Integration Differentiation (PID) control and neural network PID control were compared and analyzed, which demonstrated that the control method of neural network PID has more advantages. On the basis of the optimization, the functions of color mark cutting were developed. The actual production test of the packing machine is carried out and show that, the whole system meets the requirements of stability and packaging precision in the process of completing the corresponding functions, and has good application value.