Journal of Engineering Research and Reports

Bridge construction and maintenance represent some of the most critical challenges in civil engineering infrastructure management. With the global ageing of bridge stock and the growing complexity of structural designs, traditional monitoring methods have shown significant limitations in terms of cost, accuracy, and real-time capability. The emergence of machine learning as a powerful analytical paradigm has opened new avenues for transforming how bridges are monitored, assessed, and maintained throughout their construction and service life. This review systematically examines the evolution of bridge construction monitoring methodologies and the integration of machine learning techniques within this domain. A comprehensive synthesis of literature reveals that sensor-based monitoring systems, when combined with advanced machine learning algorithms, offer unprecedented capabilities in damage detection, load assessment, predictive maintenance, and real-time structural health evaluation. Supervised learning methods, including support vector machines and artificial neural networks, have demonstrated strong performance in damage classification tasks, whilst unsupervised approaches have proven valuable for anomaly detection in the absence of labelled training data. Deep learning architectures, particularly convolutional neural networks and long short-term memory networks, have further advanced the state of the art by enabling feature extraction directly from raw monitoring data without the need for manual feature engineering. The review also examines emerging applications in computer vision-based crack detection, acoustic emission monitoring, and digital twin integration. Key challenges including data scarcity, sensor fusion, model interpretability, environmental normalisation, and generalisation across diverse structural typologies are identified and discussed. The article concludes with recommendations for future research and acknowledges the limitations of the current review.