Integration of Stay Load Cell Measurements and Data Normalization for Structural Health Monitoring of Cable-Stayed Bridges

This study presents an integrated and data driven monitoring framework for a cable-stayed bridge that integrates traditional accelerometric measurements with load cell data to achieve a comprehensive structural assessment. Accelerometers, load cells, a weather station, and ther-mocouples are integrated into the monitoring system to capture both structural and environ-mental parameters affecting the dynamic behaviour of the bridge. The work investigates how different data normalization approaches and environmental conditions affect the accuracy of damage detection. After a procedure of modal tracking, advanced methods for data normaliza-tion are applied, including neural network-based normalization procedures aimed at reducing the effects of temperature, wind, and traffic fluctuations. The normalized outputs serve as inputs to Hotelling control charts that are analysed to assess the sensitivity, stability, and reliability of each normalization approach. A comparative study is also conducted between control charts derived from load cell based modal frequencies and those obtained from accelerometric data, assessing their relative diagnostic capability under varying environmental conditions. Results show that the different normalization strategies significantly affect control chart robustness and their ability to discern environmental effects from potential structural anomalies. The proposed methodology presents a reliable basis for long term monitoring of cable-stayed bridges by us-ing a limited but efficient sensor network.