Abstract: In the process of finishing machining of large-pitch screw, the radial depth of cut is large, the cutting edge involved in cutting is long, and the feed speed is large, which increases the cutting force by dozens of times, causing severe tool vibration and unstable thermal coupling field. In addition to tool wear, tool vibration and wear are the main reasons for the deterioration of the surface quality of the workpiece. Through the research of cutting optimization, the problem of poor surface quality of the workpiece can be solved. First, compare the fitting errors of machine learning and regression methods, and choose a more accurate machine learning method to establish a prediction model of cutting force and cutting temperature. Secondly, different swarm intelligence algorithms are used to solve the optimization goal, the solution performance of different algorithms is compared, and the optimization result of the artificial bee colony algorithm is selected as the optimal parameter combination. Finally, the surface roughness of the workpiece obtained by different cutting schemes is measured. The results show that the surface roughness of the workpiece obtained by the optimized parameter processing is reduced by 20%, which improves the surface quality of the workpiece and achieves the goal of cutting optimization.