Physics-informed reinforcement learning optimization of nuclear assembly design

Radaideh, Majdi I.; Wolverton, Isaac; Joseph, Joshua; Tusar, James J.; Otgonbaatar, Uuganbayar; Roy, Nicholas; Forget, Benoit; Shirvan, Koroush

doi:10.1016/j.nucengdes.2020.110966

Cited by 48 publications

(13 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This randomness and usage of multiple algorithms ensures independence, flexibility, and better coverage of the search space. NEORL (NeuroEvolution Optimization with Reinforcement Learning) is a set of implementations of hybrid algorithms combining neural networks and evolutionary computation based on a wide range of machine learning and evolutionary intelligence architectures [26,27]. NEORL has been developed by one of the authors of this current study, and it offers a robust implementation for ES, GWO, MFO, and DE, and so we utilize it in this work.…”

Section: Evolutionary Optimizationmentioning

confidence: 99%

Model Calibration of the Liquid Mercury Spallation Target using Evolutionary Neural Networks and Sparse Polynomial Expansions

Radaideh,

Tran,

Lin

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

The mercury constitutive model predicting the strain and stress in the target vessel plays a central role in improving the lifetime prediction and future target designs of the mercury targets at the Spallation Neutron Source (SNS). We leverage the experiment strain data collected over multiple years to improve the mercury constitutive model through a combination of large scale simulations of the target behavior and the use of machine learning tools for parameter estimation. We present two interdisciplinary approaches for surrogate-based model calibration of expensive simulations using evolutionary neural networks and sparse polynomial expansions. The experiments and results of the two methods show a very good agreement for the solid mechanics simulation of the mercury spallation target. The proposed methods are used to calibrate the tensile cutoff threshold, mercury density, and mercury speed of sound during intense proton pulse experiments. Using strain experimental data from the mercury target sensors, the newly calibrated simulations achieve 7% average improvement on the signal prediction accuracy and 8% reduction in mean absolute error compared to previously reported reference parameters. Some individual sensors experienced up to 30% improvement in their signal prediction accuracy using the calibrated parameters. The proposed calibrated simulations can significantly aid in fatigue analysis to estimate the mercury target lifetime and integrity, which reduces abrupt target failure and saves tremendous amount of costs. However, an important conclusion from this work points out to a deficiency in the current constitutive model based on the equation of state in capturing the full physics of the spallation reaction. Given that some of calibrated parameters that show a good agreement with the experimental data can be nonphysical mercury properties, we need a more advanced two-phase flow model to capture bubble dynamics and mercury cavitation.

show abstract

Section: Evolutionary Optimizationmentioning

confidence: 99%

Model Calibration of the Liquid Mercury Spallation Target using Evolutionary Neural Networks and Sparse Polynomial Expansions

Radaideh,

Tran,

Lin

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Currently, to preserve a generic implementation, the user can use variety of priori methods to convert multiobjective to single objective problem. Indeed, NEORL demonstrated successful results with complex multiobjective optimization problems, which are illustrated here [34,55,44], where -constrained and linear scalarization in conjunction with different neural, evolutionary, and neuroevolution algorithms have been effective in solving the multiobjective optimization.…”

Section: Multiobjective Optimization and Constrained Handlingmentioning

confidence: 99%

“…This inspiration has led to the term "learning to optimize" [31], which has been followed by many attempts to use RL and neural networks to solve optimization problems. See for example [32] on using RL with recurrent neural network policy for combinatorial optimization, or [33] for using graph embedding and RL to solve combinatorial optimization over graphs, or [34] for using deep Q learning and proximal policy optimization for physics-informed optimization of nuclear fuel. A comprehensive overview of using machine learning and neural networks to solve combinatorial optimization is conducted by [35].…”

Section: Introductionmentioning

confidence: 99%

NEORL: NeuroEvolution Optimization with Reinforcement Learning

Radaideh¹,

Katelin²,

Seurin³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

We present an open-source Python framework for NeuroEvolution Optimization with Reinforcement Learning (NEORL) developed at the Massachusetts Institute of Technology. NEORL offers a global optimization interface of state-of-the-art algorithms in the field of evolutionary computation, neural networks through reinforcement learning, and hybrid neuroevolution algorithms. NEORL features diverse set of algorithms, user-friendly interface, parallel computing support, automatic hyperparameter tuning, detailed documentation, and demonstration of applications in mathematical and real-world engineering optimization. NEORL encompasses various optimization problems from combinatorial, continuous, mixed discrete/continuous, to high-dimensional, expensive, and constrained engineering optimization. NEORL is tested in variety of engineering applications relevant to low carbon energy research in addressing solutions to climate change. The examples include nuclear reactor control and fuel cell power production. The results demonstrate NEORL competitiveness against other algorithms and optimization frameworks in the literature, and a potential tool to solve large-scale optimization problems. More examples and benchmarking of NEORL can be found here: https://neorl.readthedocs.io/en/latest/index.html

show abstract

“…This motivates creation of fast running accurate surrogate models for fuel performance analysis to achieve a tighter coupling with current core simulators. Particularly, we plan to integrate the surrogate model in a recent physics-informed reinforcement learning (RL) optimization framework [3] [4]. The current constraints and objectives employed in this physics-informed RL framework accommodates only coupled neutronics-thermal-hydraulics response from a commercial licensed code.…”

Section: Introductionmentioning

confidence: 99%