Artificial Neural Networks for Predicting Natural Frequencies of Concrete Gravity Dams: A Moroccan Case Study

Predicting the seismic behavior of concrete gravity dams is a critical challenge in earthquake engineering. This Study investigates the potential of Artificial Neural Networks (ANNs) in predicting the natural frequency of concrete gravity dams based on their geometric and mechanical properties. A dataset of 320 numerical simulations was developed to train and evaluate different artificial neural network architectures. The results indicate that simple neural networks with one or two hidden layers provide strong predictive capabilities for predicting the fundamental frequency, depending on the number of neurons in each layer. However, the proposed approach does not yet incorporate all the factors that may influence the seismic response of a dam, such as hydrodynamic forces and realistic seismic input. Future research could integrate nonlinear modeling and realistic earthquake excitations to validate and enhance the trends identified in this Study. These findings underscore the potential of data-driven modeling, such as neural networks, to evaluate seismic vulnerability, especially for massive concrete structures.