Abstract: The determination of process parameters for preparation of copper/graphene composites based on traditional methods is time-consuming and labor-intensive. Instead of traditional methods, machine learning models were introduced,known process parameters were used to quickly predict the friction and wear properties of copper/graphene composites.Firstly, 56 sets of friction coefficient and wear rate data were obtained by changing the content of graphene, ball milling time,and experimental load; secondly, the prediction models for friction coefficient and wear rate of copper/graphene composites were established using generalized regression neural network, back propagation neural network, support vector regression machine and random forest methods; finally, the performance of the models was evaluated. The research results show that the determination coefficient of the generalized neural network prediction model of friction coefficient reaches 93%, and the prediction model determination coefficient of wear rate is as high as 88%. Compared with the other three models, the generalized regression neural network model has higher accuracy. It provides a reference for predicting the friction and wear properties of copper/graphene composites.