Transfer learning driven design optimization for inertial confinement fusion

@article{Humbird2022TransferLD,
  title={Transfer learning driven design optimization for inertial confinement fusion},
  author={Kelli D. Humbird and Jayson Luc Peterson},
  journal={ArXiv},
  year={2022},
  volume={abs/2205.13519}
}
Transfer learning is a promising approach to creating predictive models that incorporate simulation and experimental data into a common framework. In this technique, a neural network is first trained on a large database of simulations, then partially retrained on sparse sets of experimental data to adjust predictions to be more consistent with reality. Previously, this technique has been used to create predictive models of Omega 1 and NIF 2,3 inertial confinement fusion (ICF) experiments that are… 

Figures from this paper

References

SHOWING 1-10 OF 24 REFERENCES

Cognitive simulation models for inertial confinement fusion: Combining simulation and experimental data

TLDR
This work describes a cognitive simulation method for combining simulation and experimental data into a common, predictive model, and discusses how the methods might be used to carry out a data-driven experimental campaign to optimize performance.

Transfer Learning as a Tool for Reducing Simulation Bias: Application to Inertial Confinement Fusion

TLDR
Numerical tests are carried out to investigate the applicability of transfer learning to predict the observable outcomes of inertial confinement fusion experiments under new conditions and demonstrate that an accurate predictive model can be built by retraining an initial surrogate model with experimental data volumes so small that they are relevant to the ICF problem.

A real-time machine learning-based disruption predictor in DIII-D

TLDR
By uncovering the causes of the disruption events, a better understanding of disruption dynamics is achieved, and a clear path toward the design of disruption avoidance strategies can be provided.

Machine learning for disruption warnings on Alcator C-Mod, DIII-D, and EAST

TLDR
This paper reports on disruption prediction using a shallow machine learning method known as a random forest, trained on large databases containing only plasma parameters that are available in real-time on Alcator C-Mod, DIII-D, and EAST, finding that the prediction algorithms differ substantially in performance among the three machines, but have similar disruption detection rates on a shot-by-shot basis.

Images of the gold bubble feature in NIF Gas-Filled Ignition Hohlraums

The ViewFactor experiments at the National Ignition Facility use a truncated hohlraum to allow excellent diagnostic views of the interior of the hohlraum. Time-integrated, hard x-ray (3-5 keV) images

Predicting disruptive instabilities in controlled fusion plasmas through deep learning

TLDR
Using data from plasma-based tokamak nuclear reactors in the US and Europe, a machine-learning approach based on deep neural networks is taught to forecast disruptions, even those in machines on which the algorithm was not trained.

Design and simulations of indirect drive ignition targets for NIF

Studies on simulation and design of ignition targets for the National Ignition Facility (NIF) are described. Recent effort has emphasized the systematic exploration of the parameter space of possible

Toward a burning plasma state using diamond ablator inertially confined fusion (ICF) implosions on the National Ignition Facility (NIF)

Producing a burning plasma in the laboratory has been a long-standing milestone for the plasma physics community. A burning plasma is a state where alpha particle deposition from deuterium–tritium

Review of the National Ignition Campaign 2009-2012

The National Ignition Campaign (NIC) was a multi-institution effort established under the National Nuclear Security Administration of DOE in 2005, prior to the completion of the National Ignition