Aim

Development of modular digital twin framework for operation of autonomous systems.

Objectives

1. Develop methods for splitting digital twins of autonomous systems into sets of smaller but linked-together modules that can be run independently.
2. Develop Bayesian Networks for the probabilistic modelling of dependencies between these modules
3. Assess whether these approaches can be used to speed up modelling for real-time analysis.

Description

Autonomous systems often have to operate in contested environments, so there is a desire to use digital twin models in their operation. This requires robust real time analysis across a broad range of inputs and with a comprehensive understanding of the uncertainty in the output to inform decision-making. DSTL’s definition of a DT provides a good guide to the technical requirements of a DT for this purpose. Single monolithic models lack the tractability, transparency and uncertainty treatment required for such applications and are too slow for real time analysis. This project will develop a modular approach with modules corresponding to the individual engineering components of the autonomous system. Information will be passed between modules using a probabilistic framework (a Bayesian network) so that uncertainties can be tracked through the system, making use of efficient updating to enable quick real-time analysis and a thorough understanding of what the model can and cannot tell us about the underlying system given the data available.

References

V. Volodina, N. Sonenberg, J. Q. Smith, P. G. Challenor, C. J. Dent and H. P. Wynn, "Propagating uncertainty in a network of energy models," 2022 17th International Conference on Probabilistic Methods Applied to Power Systems (PMAPS), Manchester, United Kingdom, 2022, pp. 1-6, doi: 10.1109/PMAPS53380.2022.9810635.

A.L. Wilson, C.J. Dent, M. Goldstein, Quantifying uncertainty in wholesale electricity price projections using Bayesian emulation of a generation investment model, Sustainable Energy, Grids and Networks, Volume 13, 2018, Pages 42-55