By predicting the construction elevation of a cable-stayed bridge with the help of the BP neural network, it was found that the BP neural network is suitable for the prediction of bridge construction, and an accurate prediction value can be obtained. The maximum deviation of the prediction value is 41.6 mm, which is a large error. Therefore, the use of intelligent algorithms to predict the linear elevation of cable-stayed bridge construction and reduce the impact of error can provide an important reference basis for the next stage of elevation control of standing molds and guarantee the accuracy of bridge construction control.