S15: Computational mechanics under uncertainty: inference, learning and digital twins
Co-chairs:
Eky Febrianto* (University of Glasgow)
Lawrence Bull (University of Glasgow)
Fehmi Cirak (University of Cambridge)
Keywords: Bayesian inference, uncertainty quantification, data assimilation, digital twins, scientific machine learning, structural health monitoring.
ABSTRACT
Modern computational mechanics increasingly combines physics-based models with observational data from sensors, experiments and inspections. This creates new opportunities for prediction, monitoring, and decision-making, but also raises challenges from uncertain parameters and states, sparse and noisy data, model-form error, high-dimensional inverse problems, and the computational cost of repeated forward simulations.
This minisymposium will bring together researchers working on computational mechanics problems involving inference, uncertainty and data. Topics include:
· Bayesian inversion, parameter identification and data assimilation;
· uncertainty propagation, model discrepancy and probabilistic modelling;
· reduced-order and surrogate modelling;
· scientific and physics-informed machine learning;
· digital twins, structural health monitoring and related applications.
Applications may include structural and solid mechanics, civil infrastructure, manufacturing, energy systems, aerospace and related areas. The aim is to encourage exchange between researchers in computational mechanics, applied mathematics, statistics and machine learning, with particular emphasis on methods that combine physical models with data while accounting for uncertainty.