Recognition of defects in concrete infrastructure, especially in bridges, is a costly and time consuming crucial first step in the assessment of the structural integrity. Large variation in appearance of the concrete material, changing illumination and weather conditions, a variety of possible surface markings as well as the possibility for different types of defects to overlap, make it a challenging real-world task. In this work we introduce the novel COncrete DEfect BRidge IMage dataset (CODEBRIM) for multi-target classification of five commonly appearing concrete defects. We investigate and compare two reinforcement learning based metalearning approaches, MetaQNN and efficient neural architecture search, to find suitable convolutional neural network architectures for this challenging multi-class multi-target task. We show that learned architectures have fewer overall parameters in addition to yielding better multi-target accuracy in comparison to popular neural architectures from the literature evaluated in the context of our application.

This study focusses on the estimation of seismic fragility curves for all common bridge types found in modern greek motorways. At first a classification scheme is developed in order to classify the existing bridges into a sufficient number of classes. A total of 11 representative bridge classes resulted, based on the type of piers, deck, and pier-to-deck connection. Then an analytical methodology for deriving fragility curves is proposed and applied to the representative bridge models. This procedure is based on pushover analysis of the entire bridge and definition of damage states in terms of parameters of the bridge pushover curves. The procedure differentiates the way of defining damage according to the seismic energy dissipation mechanism in each bridge, i.e. bridges with yielding piers of the column type and bridges with bearings (with or without seismic links) and non-yielding piers of the wall type. The activation of the abutment-backfill system due to closure of the gap between the deck and the abutments is also taken into account. The derived fragility curves are subjected to a first calibration against empirical curves based on damage data from the US and Japan.

A multi-objective identification method for structural model updating based on modal residuals is presented. The method results in multiple Pareto optimal structural models that are consistent with the experimentally measured modal data and the modal residuals used to measure the discrepancies between the measured and model predicted modal

No abstract

Abstract:The dynamic characteristics of two representative R/C bridges on Egnatia Odos motorway in Greece are estimated based on low amplitude ambient and earthquake-induced vibrations. The present work outlines the instrumentation details, algorithms for computing modal characteristics (modal frequencies, damping ratios and modeshapes), modal-based finite element model updating methods for estimating structural parameters, and numerical results for the modal and structural dynamic characteristics of the two bridges based on ambient and earthquake induced vibrations. Transverse, bending and longitudinal modes are reliably identified and stiffness-related properties of the piers, deck and elastomeric bearings of the finite element models of the two bridges are estimated. Results provide qualitative and quantitative information on the dynamic behavior of the bridge systems and their components under low-amplitude vibrations. Modeling assumptions are discussed based on the differences in the characteristics identified from ambient and earthquake vibration measurements. The sources of the differences observed between the identified modal and structural characteristics of the bridges and those predicted by finite element models used for design are investigated and properly justified.