Under the guiding principle of improving the reliability and maintainability of existing infrastructure, the project “Digitization of Infrastructure Structures for Structural Health Monitoring” aims to develop and apply efficient methods for reliability-based digital condition assessment using data from a variety of sensors. Structural Health Monitoring (SHM) refers to the non-destructive, continuous monitoring of structures through embedded or applied sensors that provide periodic or continuous data. This data enables the evaluation of load-bearing capacity, functionality, and durability, as well as the identification of developing damages and defects at an early stage.

Computational mechanics allows for the creation of detailed numerical models, or “digital twins,” that replicate the real-world behavior of engineering systems. These models depend on parameters such as material properties, geometry, and damage levels, many of which may be uncertain or partially unknown. In practice, the lack of precise parameter value combined with limited measurement data makes it critical to update and refine these models as new information becomes available. Bayesian updating offers a systematic framework to combine prior knowledge with new measurement data to improve model accuracy, quantify uncertainties, and generate confidence intervals for the predicted behavior of structures. This facilitates informed decision-making, such as scheduling timely and efficient maintenance activities.

Despite its advantages, Bayesian updating can be computationally demanding. Numerical models may require millions of evaluations to ensure that data and prior knowledge are integrated correctly, and new data often necessitates re-running the entire computational process. For large scale SHM applications characterized by complex models and ever-growing datasets these requirements can be particularly challenging.

To overcome these challenges, the project focuses on the Method of Moments (MoM) as a computationally efficient Bayesian updating approach. The MoM combines measurements with numerical model information in a straightforward and interpretable manner, significantly reducing the computational effort compared to conventional methods. The TU Dortmund research team brings extensive expertise in Bayesian methods and the MoM framework, with proven applications in structural reliability. This knowledge will be adapted and validated for SHM scenarios, including the presence of large datasets, parameter uncertainties, and noisy measurements.

The work within this project is structured into three key elements: (1) Implementation of Bayesian updating, (2) Development of a computationally efficient methodology , (3) Validation and demonstration of the approach.

Link to DTEC-SHM