Comparison of temperature compensation methods in model updating for structural damage assessment

Vibration-based damage assessment is a widely used tool for structural health monitoring of buildings, infrastructure, and wind turbines, ensuring their safe operation. However, damage and environmental conditions, such as temperature variations, simultaneously influence the vibration behaviour of a structure. Therefore, the impact of temperature changes must be taken into account in the damage assessment process. Model updating is a promising method for damage localisation and quantification. This work presents a comparative study of two general approaches for compensating temperature variations in model updating. The first approach involves training a regression model on the damage-sensitive features of the model updating problem on the known temperatures. Subsequently, model updating is performed for damage localisation and quantification based on an approximated reference state. The second approach integrates temperature compensation directly into the model updating process. This is achieved by scaling the stiffness values of the entire structure to account for temperature variations, while modifying only specific elements for damage assessment. These approaches will be compared using a laboratory steel beam in a climate chamber and the Leibniz University Test Structure for Monitoring (LUMO). Recommendations will be provided for different application scenarios, offering insights into the strengths and potential uses.