In order to reduce the volume shrinkage of polyamide 66（PA 66） gear injection molding as an example,by using the orthogonal test and Moldflow simulation software to obtain data as training samples of neural network,the neural network prediction model of the volume shrinkage of the gear is established,and the accuracy of the artificial neural network model is tested by the sample. Therefore,the time of optimizing process parameters is shortened,and the process parameters are optimized. The results show that,the efficiency of process design can be greatly improved by using the optimization of molding process parameters of plastic gear with combining the orthogonal test,numerical simulation and artificial neural network,and the accuracy is high.