Abstract: To accurately predict the CH4 replacement rate after injecting CO2 gas into coal seams, this article takes anthracite as the research object, and based on the CH4 replacement rate η of coal samples under different coal seam temperatures T, different CO2 injection pressures P0, and different initial adsorption equilibrium pressures P1 Based on experimental data, a prediction model for the CH4 replacement rate of the coal sample was established using the BP neural network algorithm. The accuracy of the prediction model was evaluated through the determination coefficient R2, root mean square error RMSE, and mean absolute error MAE. The results indicate that the CH4 replacement rate η of the coal sample increases with the increase of coal seam temperature T and CO2 injection pressure P0, but decreases with the increase of initial adsorption equilibrium pressure P1; The established prediction model has a determination coefficient R2 value of 0.986, a root mean square error RMSE value of 0.349, a mean absolute error MAE value of 0.312 in the application of experimental data in the test set. The error between the predicted value and the actual value does not exceed 2.30%, indicating that the prediction accuracy of the established prediction model is high and can provide guidance for gas injection mining in this coal seam.