Nickel-based alloys are widely used in the manufacture of blades for aviation turbines and compressors,and improving their processing efficiency and quality is a hot spot in the industry.In order to realize the accurate prediction of the surface roughness of the constant force grinding and polishing of aero-engine blades,a prediction model based on the chaos sparrow search algorithm (CSSA) optimization of BP neural network was proposed,and the research on the prediction of the process parameters of the robot abrasive belt grinding and polishing was carried out.Four main influencing factors were selected as variables to conduct orthogonal experiments and range analysis.BP neural network was used to establish a surface roughness prediction model,and Tent chaotic map and population diversity variation were introduced to conduct the sparrow search algorithm.After improvement,the prediction model was optimized by CSSA.The results show that the roughness prediction error after optimization is reduced by 85.9%,which is obviously improved compared with that before optimization.The roughness value corresponding to the input process parameters can be accurately predicted.