In this paper, Damage Assessment(damage locations and extents) is accomplished by using neural networks. Among the neural networks, backpropagation neural network is adopted to fulfill the damage assessment. When the damages occur, the structural responses are changed and the responses contain the damage locations and extents, implicitly. Throughout the changes of the structural responses, the neural networks detect damage locations and extents. Because this scheme is to find the causes from the results, we call it the inverse process. When we use the neural networks for the damage assessments, the work finding the damage locations and extents is fulfilled simultaneously.
The conventional damage assessments have used the ideal properties, seismic loads and static loads. In such cases, it is difficult to obtain the data surveying on the structures, practically. But, it is much easier to measure the vibration signature under the moving loads. The vibration of the beam structure is induced by the moving load. The acceleration is captured. This is the input of the neural networks. The neural network is trained and tested with the acceleration data. The simulation informs that it is efficient to apply the neural network to the damage assessment.