Research On Multi-stage Scheme For Identifying Damage In Steel Truss Arch-bridge Based On Neural Network

Steel truss arch-bridge structure is an important structural form of bridge, steel truss arch bridge has majestic appearance, long span, high carrying capacity ,so it develops rapidly in recent years, such as Tianjin Cathay Pacific Bridge and Chongqing Chaotianmen Yangtze River Bridge. The Chaotianmen River Bridge has a span of 190 +552 +190 m, and it is the world's largest steel truss arch bridge. Many bridges may have different levels of damage during use, thus may lead to accidents and endanger people's lives. Therefore, damage identification of bridge structures and early warning of damages have great importance.The damage identification of bridge structures (especially the steel truss arch-bridge) has become a major civil engineering research project. The damage of the bridge will cause changes of dynamic characteristics of the bridge, so if we can establish a relationship between the changes of dynamic characteristics and structural damage, we can easily find the damage element. Neural network has strong nonlinear mapping ability and anti-interference capacity, so it is very suitable for structural damage identification.In this paper, long span steel truss arch-bridge will be used as research object to study damage identification. Using neural network technology, we propose a multi-stage damage identification method of long-span steel truss arch-bridge structure . Under the foundation of this approach, a kind of high efficient neural network methods to identify damage in arch-bridge structures has been established by writing APDL program of ANSYSand MATLAB programs.The multi-stage identification scheme for identifying damage in steel truss arch-bridge structures has been used to study identifying damage in structural members. This scheme is divided into four steps. Firstly, we use damage anomalous filter which is set up by neural network to alarm the damage in structural members. Secondly, the primary location of the member damage is determined by the neural network with inputting the combined damage index X1. At the third step, the specific location of the member damage will be determined by the neural network with inputting the combined damage index X2. Finally, the damage degree of the member is determined by neural network with inputting the change rate of squared(RNF) modal frequency.