Working group: Interdisciplinary monitoring

This category focuses on the geodetic deformation monitoring of civil infrastructure and structures using different deformation models such as congruence, kinematic, static, and dynamic. The group employs advanced geodetic and remote sensing techniques, including TLS, image-assisted total stations (IATS), laser trackers, micro-electromechanical systems (MEMS), photogrammetry, and Synthetic Aperture Radar (SAR), to detect and analyse deformations in structures and infrastructures such as buildings, bridges, dams, roads, railways, and wind energy turbines. These technologies enable the early detection of deformation and help predict maintenance needs, guiding timely interventions to ensure the preservation and safety of infrastructure.

This section focuses on exploring advanced geodetic analysis techniques to better understand deformation patterns and monitor deformations in both natural and man-made structures and infrastructures. The research emphasizes the application of sophisticated deformation analysis algorithms, alongside deep learning-based models, for the intelligent and automatic detection of structural defects.

In point cloud-based deformation analysis, the group develop B-spline/Multilevel B-spline curve- and surface-approximation methods to 2D and 3D point clouds. These approaches ensure accurate modeling of deformations, particularly in complex environments, such as underwater monitoring.

Deep learning-based anomaly detection is utilized to automatically identify deviations from normal behavior in structural systems, enabling efficient and automated analysis of large-scale deformation datasets. This enhances the robustness and reliability of monitoring systems. Additionally, the researches explore the application of deep learning models for anomaly detection in Persistent Scatterer (PS) time series data, improving the accuracy in large-scale anomaly detection.

Furthermore, this section covers the intelligent detection and 3D reconstruction of cracks in civil infrastructure from photogrammetry imagery. These techniques contribute to a more precise understanding of structural health, aiding in the identification of potential risks and necessary interventions.

This category focuses on analyzing the dynamic behavior of structures, such as bridges and wind energy turbines, to assess their structural health. The research in this area utilizes geodetic time series measurements obtained from sources like MEMS, video captured from the IATS, or TLS profile data for vibration monitoring. Advanced methods, including univariate autoregressive (AR) modeling and multivariate vector-autoregressive (VAR) modeling, are employed to estimate modal parameters such as eigenfrequencies, eigenforms, and damping ratio coefficients.

These techniques enable the monitoring of structures by estimating deterministic parameters (modal parameters), auto- and cross-correlations, and stochastic model parameters. Early detection of instability or potential failure is facilitated, enhancing the safety and resilience of the infrastructure.

The integration of diverse geodetic sensor technologies enhances modern infrastructure monitoring by combining different (contact and non-contact) measurement techniques. TLS point clouds, fused with digital photogrammetry, improve accuracy and completeness in area-based deformation analysis by optimally combining 3D point cloud and image-derived features across different epochs. To mitigate accuracy degradation in MEMS accelerometers, they are integrated with geodetic measurement systems such as IATS using Kalman filtering, ensuring more reliable absolute position estimates. For large-scale monitoring, GNSS and InSAR data fusion within a Kalman framework provides high spatio-temporal resolution, enabling precise and reliable deformation monitoring.

This research focuses on evaluating model and measurement uncertainties in TLS data for precise and accurate deformation monitoring. In addition, the group develops robust Bayesian adjustment methods for geodetic time series analysis, addressing outliers, colored noise, data gaps, and nonstationary noise.

To ensure reliability in InSAR data, an advanced spatio-temporal quality model—incorporating univariate/multivariate time series analysis, Multilevel B-Splines (MBA) for surface approximation, and bootstrapping for uncertainty quantification—enhances the reliability of InSAR data in large-scale deformation monitoring.

Spatio-temporal modelling in geospatial data is conducted by integrating univariate/multivariate time series analysis with B-spline/polynomial curve/surface approximation, depending on the data's dimensionality and the application. For example, 2D spatio-temporal models are developed using 2D TLS profile measurements, which are particularly suitable for monitoring structures such as wind turbines and bridges. In contrast, 3D spatio-temporal modelling is applied to InSAR time series data to capture spatio-temporal deformation behaviour across local or global surface patches.