Abstract Signal processing is crucial for bridge Structural Health Monitoring (SHM), mainly because bridges face constant changes in traffic loads, temperature cycles, and environmental factors along with seismic events, making it difficult to interpret their structural responses. A four-stage bibliometric process identified 201 studies focused on bridges, which were validated in Scopus and Web of Science to ensure relevance and citation quality. These studies cover a variety of bridge types and discuss different sensing setups, preprocessing methods, feature extraction techniques, and interpretive models used in practice. Traditional signal-processing methods form the foundation for modal identification, vibration-based condition assessment, and long-term tracking of stiffness changes. At the same time, learning-based frameworks are gaining the potential to predict nonlinear structural behaviors and mitigate the complex effects of temperature, traffic variability, and environmental impacts that often mask damage indicators. A keyword co-occurrence analysis highlights growing interest in fiber-optic sensing, vision-based inspection, hybrid data-fusion schemes, and ecological compensation strategies to fill gaps in monitoring. Despite these advances, ongoing challenges include limited large-scale field validation, poor interpretability of machine learning models, and a lack of standardized datasets for benchmarking algorithm performance. This review offers a broad overview of how signal-processing techniques for bridge SHM have developed from 1985 to 2025, spanning traditional analytical methods to modern data-driven and learning-based approaches. It categorizes the selected studies into a clear taxonomy linking sensing methods, signal-processing workflows, and diagnostic models. As a result, it underscores the practical strengths and limitations of traditional, data-driven, and hybrid approaches in bridge applications.
Hajivand et al. (Fri,) studied this question.