Development of an Anomaly Detection Model for Bridge Vibration Acceleration under Temperature Variations and Investigation of Thermodynamic Mechanisms

Bridges play a critical role in transportation systems, making their long-term operational safety an urgent concern.The vibration characteristics of bridges are influenced by various external factors, including temperature, load, and wind.Among these, temperature fluctuations significantly affect the dynamic response of bridge structures.Thermal expansion and contraction of bridge materials induced by temperature changes can alter vibration characteristics, impacting the accuracy of structural health monitoring.Traditional bridge health monitoring methods primarily rely on static analysis and manual inspections.However, under complex environmental conditions, particularly with significant temperature variations, these methods struggle to provide accurate and real-time monitoring results.Recently, vibration acceleration-based anomaly detection methods have garnered research attention, yet the influence of temperature on vibration characteristics remains insufficiently addressed.Current studies predominantly focus on applying machine learning and deep learning techniques to bridge vibration anomaly detection but often neglect the disruptive effects of temperature variations, limiting their adaptability and accuracy in real-world applications.To address this issue, this study analyzes bridge vibration characteristics under varying temperature conditions, constructs a temperature-inclusive vibration acceleration dataset, and proposes a novel anomaly detection model to enhance detection precision and robustness.The findings offer a new perspective for bridge health monitoring and provide theoretical insights for managing and maintaining bridges under temperature variations.