Model calibration of slender minarets based on artificial neural networks

Abstract

Model calibration plays a decisive role in the assessment of structural damage using techniques developed by model-based approaches. In this study, an artificial neural network (ANN) based procedure is presented to construct the baseline model of the 24.25 m high minaret of Hacılar mosque (Turkiye). Since the swaying behaviors of slender minarets are comparable to those of a beam-like system, a three-dimensional (3D) system formed by brick elements is transferred into a simplified system composed of only beam elements regardless of its peculiar geometries as well as the complicated connections between the booting part and surroundings. The simplified model is calibrated by targeting the modal information generated on the 3D building using a 4-uniaxial accelerometer network capturing only the lowest modes. The physical simplified model is promptly calibrated by adjusting not only stiffness but also mass and Poisson’s ratio parameters simultaneously. As a result, the modal information of the calibrated model gets closer to the counterparts obtained from the 3D system, especially in the fundamental mode.