Modern remote sensing technologies now provide the basis for flexible and highly accurate three-dimensional geometric modeling of structures in the form of point clouds. To date, most efforts are focused on how to use these point clouds to form a digital twin of an asset, but these models can also be used to augment and improve condition assessment and structural health monitoring (SHM). However, point cloud analytics require unique approaches given the complexity and scale of the data. To illustrate these capabilities, we propose a new SHM method that leverages 3D point cloud data and the evolution of this data over time. Taking inspiration from recent work on the use of complexity measures for sensor driven SHM, here we adapt the concept for spatial analysis of 3D digital twins. The fundamental assumption that underpins the approach presented here is that, as a structure degrades in integrity, the randomness of the data increases when compared against the null model of the homogeneous Poisson process, otherwise described as ‘complete spatial randomness’ (CSR). In spatial point analysis, points from a baseline model are generated and placed within a normalized Cartesian reference frame. The spatial randomness of this baseline is considered the null model of the homogeneous Poisson process. In subsequent 3D models of an asset, spatial complexity metrics are recomputed on a local neighborhood level, with increased complexity corresponding to potential damage or degradation of the asset. Another question of interest is to provide a suitable mathematical model for this underlying temporal evolution. Compared to more conventional analytical approaches that can only detect data anomalies via a single computation, this complexity-based approach enables us to further integrate multi-level information, in the form of first and second order moment metrics, to evaluate data anomalies in more depth. In this method we use the variation of the first and second moments of the average intensity of the points in space. A first order metric of a point pattern represents the density change across the study region such as Quadrat density or Kernel density. The second-order metric of the point pattern considers the distance between points, effectively quantifying how points are distributed relative to one another. Examples include Ripley’s K-function, the L-function or Baddeley’s J-function. This analytical approach was tested on a variety of laboratory scale specimens with varying levels of damage and degradation. The results show that this new technique provides rapid analytical capabilities for finding damage and quantifying both damage and evolution in point clouds. Ongoing work seeks to scale up these measures to full-scale specimens, and to explore methods of using the results for damage prognosis through statistical time-series modeling of the evolution of the complexity metrics.

This content is only available via PDF.
You do not currently have access to this content.