The work aims to improve the coil edge uniformity, improve the quality of the packaging products and reduce the packaging material loss. The cause for coil deviation in the unwinding process was analyzed. Based on the analysis of the mathematical model for the unwinding deviation control system, a deviation controller based on the fuzzy neural network PID was proposed to achieve the complicated parameter self adjustment and optimization of the coil unwinding system. The simulation and experimental results showed that, the fuzzy neural network PID had faster response speed and smaller overshoot, and the deviation control accuracy was up to ±0.5 mm. The proposed control method can obviously reduce the deviation error of the flexible packaging material, and greatly improve the packaging quality of the product.