Optical fibre sensors and deep learning algorithms based structural health monitoring framework for reinforced concrete beams

Optical fibre sensors and deep learning algorithms based structural health monitoring framework for reinforced concrete beams

This thesis presents a novel framework for structural health monitoring (SHM) of reinforced concrete (RC) beams based on fibre optic sensor (FOS) technology, finite element analysis (FEA), and deep learning (DL). The proposed research addresses the limitations of existing SHM methods by constructing a tailored sensor network, a comprehensive strain dataset, and an efficient DL model for accurate predictions. The study began with a comprehensive analysis of current SHM practises, focusing on applying and incorporating FOS, FEA, and DL in monitoring structural health. Distributed optical fibre sensors were used to establish a sensor network and acquire strain data from RC beams subjected to different loading conditions. Concrete damaged plasticity-based FEA model was established and validated with experimental strain data. The validated model has been used to generate a strain dataset. This dataset was then used to train a DL model for predicting the structural health of RC beams based on artificial neural network architecture. The proposed SHM framework was exhaustively validated via a two-tiered experimental procedure involving short and long-span RC beams subjected to various loading scenarios. The predictive capabilities of the DL model were evaluated rigorously using the extensive strain data derived from these experiments. The model prediction has been classified into eight classes, and the prediction accuracy was impressive 81.25%. Sensitivity analysis revealed a robust prediction accuracy of 74% with only 20% of input data. This study is novel due to its integrated approach to SHM, which leverages the assets of FOS, FEA, and DL to provide precise, data-driven insights into the structural health of RC beams. This method not only improves the efficacy and precision of SHM, but it also has the potential to be applied to other types of structures, thereby creating new research opportunities and field advancements.

File reproduced in accordance with the copyright policy of the publisher/author/creator.