Identification of damaged members and damage extent in bridge deck by neural network

Determination of a damaged structural member and its damage extent is the last step of structural damage identification, and it is also the fundamental work for further decision making for structural safety. A method for identifying the damaged member and damage extent simultaneously by a back-propagation neural network is investigated. The training data construction and training strategy for the network are proposed. By taking the cable-stayed Kap Shui Mun bridge as an example, the method is demonstrated. On the basis of sensitivity analysis of modal parameters to damage, 12 natural frequencies and 6 components of mode shapes are selected as the basic data to configure the input vector of the network. The output vector of the network is the indicator of both damaged members and damage extent. When the damage extent is larger than 60%, 9 of 10 cases simulated are identified correctly for damaged member and more than half of cases are quantified acceptably for damage extent.