Abstract: Taking a factory helical cylindrical gear reducer as the research object, a simulation model of the reducer is established by Romax Designer software, and the optimal gear trimming parameters of the reducer are searched by combining genetic algorithm, aiming to improve the overall smoothness of the reducer and reduce vibration and noise. Combined with the theory of gear reshaping optimization, the optimized parameters are tooth slope, tooth bulge, involute slope, involute bulge and tooth top trim edge, and the optimized targets are gear transmission error, peak load on tooth surface and tooth surface load distribution, and finally the optimized data obtained by genetic algorithm are applied to the reducer model. Comparing the simulation results before and after shaping, it can be seen that the gear transmission error after shaping is greatly reduced, the peak load and stress on the tooth surface are greatly reduced, the tooth surface bias load problem and noise are greatly improved, and the life and transmission smoothness of the reducer are effectively improved.