The relationships among the process parameters and material removal rate in magnesium alloy chemical mechanical polishing (CMP) process are nonlinear,which leads to the difficulty to establish a comprehensive and accurate model to predict the material removal rate.A model for material removal rate in magnesium alloy CMP process based on neural network (NN) optimized by genetic algorithm (GA) was proposed,in which the typical data were obtained from the orthogonal experiments of magnesium alloy CMP.The model was applied to investigate the influences of input parameters involving the polishing pressure,the polishing head speed,the slurry flowing rate and polishing time on the material removal rate.The performance of the model was verified by simulation and the experimental data.It is shown that the data of GANN forecast are in good agreement with the experimental data,and the fitting degree of GANN model is in the range of 9322%~9797%,which shows a much better stable ability than that of NN model in the range of 7156%~9356%.The results show that the GANN model can be used to predict the material removal rate in magnesium alloy CMP process,it can meet with the industrial requirements.