In this paper, using BP neural network theory, control parameter prediction research has been performed combined with a cable replacement project of a double pylon concrete cable-stayed bridge with three-span. Making construction control in phases, the theoretical cable force and elevation at construction state can be worked out by finite element theory. Then the predicted values of the cable force and elevation at various construction states would be received, through building simulation model of neural network by MATLAB software system. The predicted values can provide some references for cable replacement project. Through comparing the predicted values and the measured values of the cable tension and elevation at various construction states, the predicted values and desired variable were similar. And the changing trend of them was accordant. It proved the efficiency and feasibility of the construction control parameter prediction using BP neural network theory.