Graphnarx model predicts structural motion using node interactions
Towards a GraphNARX model for structural dynamics
Machine Learning
Summary
Structures like buildings or bridges move and shake over time due to forces acting on them, and predicting this behavior helps keep them safe. The authors introduce GraphNARX, a model that uses graphs to represent structural parts and how they influence each other. This model looks at past movements and forces at different points and predicts the next movement. Because it learns locally between connected parts, it can also guess movements where measurements are missing, acting like virtual sensors. Their tests show the model predicts well and works for structures of different sizes.
What this means in practice
- •For structural engineers: Predict structural responses to forces for monitoring and safety using a graph-based model that adapts to various building sizes.
- •For infrastructure maintenance teams: Use virtual sensing to estimate displacements at unmeasured points on structures, reducing the need for extensive physical sensors.
Authors
George Tsialiamanis, Keith Worden
Abstract
The current paper introduces the GraphNARX model, a graph-based nonlinear autoregressive exogenous model for structural dynamics. The model is built to account for the geometry of structures and to model dynamic behaviour as the interaction between elements of the structure. Similar to other types of NARX models, GraphNARX is used for system identification. Nodal attributes of the graph encode past displacements and nodal forces, and the output of the model is the next-timestep displacement. The graph formulation of the model allows it to generalise to structures of different dimensions and degrees of freedom. Moreover, because graph-learning is performed locally between the nodes, the GraphNARX model can also learn to infer displacements of nodes for which data are not available, i.e.\ to perform virtual sensing. The proposed model's performance is evaluated in a system-identification framework and a virtual-sensing framework, and achieves good accuracy and exhibits generalisation capabilities to structures of different sizes.