Characterization of Relative Movements between Blocks Observed in a Concrete Dam and Definition of Thresholds for Novelty Identification Based on Machine Learning Models

Dam surveillance activities are based on observing the structural behaviour and interpreting past supported by knowledge of main loads. For day-to-day activities, data-driven models usually adopted. Most applications consider regression for analysis horizontal displacements recorded in pendulums. Traditional not commonly applied to relative movements between blocks due non-linearities related simultaneity hydrostatic thermal effects. A new application a multilayer perceptron neural network model is proposed interpret measured hourly concrete dam under exploitation. methodology threshold definition novelty identification, taking into account evolution records over time responses study. The results obtained through case study showed ability presented this work characterize movement identification novelties behaviour.