Combining Numerical Simulations, Artificial Intelligence and Intelligent Sampling Algorithms to Build Surrogate Models and Calculate the Probability of Failure of Urban Tunnels

When it is necessary to evaluate, with a probabilistic approach, the interaction of urban tunnels neighboring structures, computational power an important challenge for numerical models. Thus, intelligent sampling algorithms can be allies in obtaining better knowledge result’s domain, even if possession smaller number samples. In any case, when limited, evaluation risks also restricted. this context, artificial intelligence (AI) fill gap risk analysis by interpolating results and generating larger samples quickly. The goal AI algorithm find approximation function (also called surrogate model) that reproduces original simulation behavior evaluated much faster. This constructed performing multiple simulations at special points obtained techniques. paper used hypothetical case validate methodological proposal. It concerns sequential excavation tunnel, about three diameters deep, interacting seven-story building. First, three-dimensional model (FEM) was solved deterministically, then its domain mesh were refined. After that, another 170 solutions numerically from FEM software, strategically random variables involved. Sequentially, based on 31 techniques, which greatest importance predicting magnitude vertical displacement foundation elements surrounding Then, once most selected, techniques again trained tested define one least R-squared. Finally, using best-fit algorithm, possible calculate probability failure massive samples, sizes order 107. These illustrate convergence Simple Monte Carlo Sampling (MC) Latin Hypercube (LHS). main contribution methodological; therefore, new procedure aggregated state-of-the-art assessment methodologies tunnel-related problems.