To study the nonlinear correlation between strata parameters, shield tunneling parameters and ground settlement, an ABC-BP neural network model that can predict the ground settlement was established by using the artificial bee colony (ABC) algorithm and BP neural network based on the shield interval of Nanjing Metro Line 6. This model was validated through three consecutive sections of the shield. The results show that the prediction accuracy and prediction stability of ABC-BP model are better than BP model, and the predicted value is consistent with the real value. The model can accurately reflect the evolution law of ground deformation during the shield machine approaching the monitoring section and achieve the purpose of final ground deformation control. This study proposes the on-site application concept of ABC-BP neural network, constructs the relationship between strata parameters, excavation parameters and settlement, and can directly control shield tunneling parameters and ground deformation through strata parameters.