2025/11/28 by Möller, Sören, Jones, Matthew R., Jonscher, Clemens +3
#600 | Technik::620 | Ingenieurwissenschaften und Maschinenbau #EOVs #Gaussian process regression #Grey-box modelling #Mode shapes #Model-based SHM
paper · doi:10.15488/20114
Structural health monitoring (SHM) offers a promising path towards automated, long-term monitoring of critical infrastructure such as offshore structures and bridges. A crucial component of such monitoring schemes is the localisation of damage, with model-based SHM providing a possible framework for this task. Here, damage localisation can be achieved by minimising the discrepancy between the modal properties of a damaged state and those of a healthy reference, but it is severely hindered by changes in the modal properties induced by environmental and operational variations (EOVs). This difficulty is compounded by the fact that measurement data are often limited, with data commonly unavailable across the full operational span, missing due to sensor failure and dropout, or sparsely sampled because of hardware constraints. In particular, mode shapes are often used without adequately accounting for the limited coverage of EOVs in continuous model-based damage localisation frameworks, resulting in inaccurate localisation. In this paper, we propose a regression-based data normalisation scheme that learns vector-valued mode shapes explicitly as functions of EOVs, allowing them to vary across a structure’s operating envelope and thereby enabling model-based damage localisation under varying environmental and operational conditions. To further alleviate the problem of limited training data, we consider two grey-box modelling strategies based on Gaussian processes that incorporate accessible engineering knowledge: (i) a Hilbert-space Gaussian process enforcing boundary conditions, and (ii) a Gaussian process with a physics-based prior mean given by finite-element mode shapes. We target long-term monitoring scenarios, where training data are limited and standard machine learning regression techniques underperform. Using the Leibniz University Test Structure for Monitoring, a representative real-world structure, we show that incorporating accessible engineering domain knowledge into a Gaussian process alleviates data scarcity and improves damage localisation under partially observed EOVs. This significantly extends the range of scenarios where model-based damage localisation is feasible, thereby improving the continuous, long-term monitoring capabilities of operational structures.