Assessing the seismic sensitivity of bridge structures by developing fragility curves with ANN and LSTM integration

In today’s transportation networks, bridges play an essential role as conduits that allow efficient access to a variety of locations. These structures are still vulnerable to outside pressures, though, and doing so can result in serious harm, especially during seismic occurrences. In this research, we model and analyze reinforced concrete (RC) T-beam bridges with elastomeric bridge bearings in order to thoroughly assess the seismic behavior of bridge components. We build and examine several span bridge models with CSI Bridge Software, altering pier heights and bearing stiffnesses in a methodical manner. In this work, we evaluate an RC bridge’s seismic susceptibility by taking regionally variable ground motions into account. Fragility curves, which are crucial instruments for evaluating risk, are at the center of our research. The probability of failure is represented by these curves over the whole load spectrum. Typically, fragility curves plot estimated probabilities (such as deflection) against ground motion parameters, providing insights into the likelihood of exceeding specific deformation limits during seismic events. Our research aims to create accurate fragility curves, facilitating precise loss calculations for bridge structures. By employing artificial neural networks (ANNs) and long short-term memory (LSTM), this research addresses uncertainties associated with influencing factors. It has been discovered that the inputs and outputs of the ANN and LSTM models are, respectively, the influencing traits and fragility parameters of significant components.