The study presents the data-driven applications of continuous wavelet transforms (CWT), orbit-shaped analysis and convolutional neural networks (CNN) using GoogLeNet for Dębica railway bridge health monitoring in Poland. Training and validation data sets are the dynamic behavior of the bridge deck recorded through an IEPE vibration sensor with a sampling frequency of 128 Hz from vibration-based structural health monitoring (SHM) system over a nine-month period from December 2019 to September 2020. Utilizing Morse, Morlet, and Bump wavelet, the vibration signal scalogram images are produced in the frequency–time domain as the input for CNN classification models, while the output is to predict health states based on the experimental tension force of eight hangers using label thresholds developed by calibrated finite element model. Moreover, the vibration-based orbit-shaped image patterns, acquired through a bidirectional sensor on each hanger are processed with CNN classification models for automated hanger health diagnostic. The accuracies and weighted F1-scores of the wavelet-assisted and orbit-shaped CNN models are more than 83% on imbalanced validation images. The results show that the proposed CNN approaches can classify the potential structural problems.

Diagnostic load testing refers to the use of the measured historical responses of the structure in the field data to better understand its dynamic and static structural behaviours. It is important and necessary to predict the health state, load capacity, and aging of the structure by updating the finite element (FE) model, which can give useful information to aid the design of retrofits and the maintenance of the existing bridge in the future. The paper presents an update of the full-scale FE model for the reinforced concrete (RC) bridge structure over the seawater river based on the experimental strains under the static load testing in which the representative FE model of the actual structure is determined from the optimisation procedures. The optimisation variables are applied, including the cross-sectional properties and concrete material calibrated through the genetic algorithm (GA) optimisation in the MATLAB software, which interfaces with the FE modelling in the scripting of the SOFISTIK TEDDY software automatically. The bending moments at the mid-span of the RC girders are determined in the FE modelling to compute stresses, which are compared with the measured stresses through optimisation scenarios with a percentage error of the objective function less than 10%. The measured data of concrete strains are recorded from reusable strain transducers installed on the mid-span girders for every bridge span, which are used to calibrate the bridge model in static load testing. The novelty of the solution is to implement innovative techniques using field data as an improved approach for calibrating automatically the analytical FE model parameters of all RC spans of the bridge until its static behaviours are very similar to those of the actual bridge. The final updated FE modelling is used to apply truck load configurations according to bridge design standards such as the AASHTO specifications, which can predict the load limits of the existing bridge structure more accurately and reliably. These proposed approaches can be applied to large bridges as well as complex structures with supporting FE analysis software and data processing software.

The paper studies estimate ultimate strength and local buckling of circular concrete filled steel tube (CFST) specimens subjected to axial load using strain gage, force transducer and displacement sensor. The strain responses of CFST column in testing procedure and the results of finite element modelling will be used to predict failure modes when cracks occur in concrete core and deformation increases on steel tube's surface. Besides, the principles of Eurocode 4 and AISC 360-10 are used to calculate the critical buckling stress in comparison with experimental approach and numerical evaluation. Combining experimental and simulation data can be very useful for diagnostic load rating and limit, and effective to evaluate the damage on structural health monitoring of CFST members.