Research Of The Improved Nerual Netwoeks On Parameter Iddentification Of Cable-stayed Bridges

Cable-stayed bridge construction control is a complex engineering system, the actural value of the structural parameters are always differed from the designed ones. Because of the existence of these difference, the configuration and internal force of the constructed bridge will be biased with the designed state. So an accurate parameter state is necessary for construction control. However, some parameters cannot be obtained directly through the existing measuring methods. So based on the engineering background of Jiashao Bridge, the improved neural networks are applied in parameter identification of cable-stayed bridge, the main works are as follows:(1) The specialties of the least square method, Calman filtering method, grey system theory and the theory of artificial neural network which had been successfully applied in the construction control are respectively analyzed. Then the superiorities of the neural network in multi-parameter identification are explored.(2) Based on the practical engineering of the Jiashao bridge, the swatches for neural networks are obtained by theoretical calculation of FCM software-NLABS. And the disadvantages of traditional BP network are analyzesed from the results.Corresponding optimized methods are discussesed.(3) Compared with the traditional BP network,the accuracy and reliability of the improved network are discussed through the comparative analysis of recognition errors. The feasibility and superiority of genetic algorithm based on ANN are anasysed.(4) To apply the genetic algorithm based on ANN with the Jiashao bridge and the results show that the optimized neural network recognition results conform to the objective laws, the result proves its feasibility in the multi-parameter identification of cable-stayed bridge.