The Application Of BP Neural Network In Cable Replacement Of Cable-Stayed Bridge

Basing on the collection and analysis of a large number of data and theories about cable replacement and application of neural network, for cable-stayed bridge is a high order statically indeterminate system, and the distribution of the cable is different from each other, this paper put forward new methods for construction control of cable changing for cable-stayed bridge, test of cable tension and determining the cable stresses during construction with BP neural network technology as the foundation. It is difficult to the utilize traditional methods to set up models and reflect the change laws of the bridge because of the structural characteristics of cable-stayed bridge, but the artificial neural network has obvious advantages in solving the such issues.Basing on project of Qianwei Min River Bridge, using finite element analysis software MIDAS as a tool, this paper extract the learning samples for the study of the network and test samples for the test of the network .In accordance with their respective characteristics, the paper chooses network type, network topology, transfer function of each layer, characteristic parameters, data pre-processing methods and the number of training samples for each model of artificial neural network accordingly. With MATLAB neural network toolbox and using BP algorithm, the paper construct neural network models used for construction control of cable changing for cable-stayed bridge, test of cable tension and determining the cable stresses during construction. For the first two models, the paper completes the input of the data, as well as training the networks. Finally inputting the input data of the test samples to the network, getting simulation output, then comparing the simulation value with output data of test samples, and testing the effect of the training, the results are satisfactory.Study shows that artificial neural network has a strong feasibility and reliability, has significant advantages in solving the nonlinear problems and has broad application prospects.