The purpose of this work is to study the error of the three-loop PID (Packet Identifier) control model of cradle CNC (Computer Numerical Control) machine tool in the cutting process through simulation. Here, the milling and hobbing process are controlled through the vertical feed servo system of the cradle CNC machine tool. At the same time, the model is established in the ADAMS (Automatic Dynamic Analysis of Mechanical Systems) and MATLAB (Matrix Laboratory) software, including the servo motor, the complete control system, and the mechanical operation system. Then, the parameters of the three-loop PID are re-tuned. The gain parameters are adjusted from small to large, and the integral time constant is adjusted from large to small following the general experience of the external load. The steady-state value without vibration overshoot is set as the optimal value. Finally, the cradle CNC machine tool and the hobbing model are constructed, and the milling and hobbing are simulated through experiment. The pitch deviation and accumulation deviation of the simulated gear are measured through the VGMC (Virtual Gear Measurement Centre). The simulation experiment shows that the initial values set through the PID controller can shorten the system instability that arose from random values and shorten the intelligent algorithm and optimization time. Therefore, the error of the cradle CNC machine tool is significantly smaller than that of the ordinary CNC machine tool in the gear cutting process when controlled by a servo drive system with more accuracy, rapidity, and stability. It provides a more effective method to reduce the error in future precision machining. This can improve the accuracy requirements for products in the machining of CNC machine tools in China and lays a foundation for future intelligent control research.