A novel irradiation module for ANICCA fuel cycle code based on multi-task learning
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
This paper introduces an alternative approach to irradiation modelling within the context of nuclear fuel cycle codes like ANICCA, the fuel cycle code developed by the Belgian Nuclear Research Centre (SCK CEN). The focus lies on upgrading the irradiation module by substituting the CRAM-based depletion calculations for a more flexible and innovative approach based on Multi-Task Learning (MTL). Utilizing data generated with SERPENT2 Monte Carlo simulations, two MTL neural networks are developed to surrogate irradiation processes of Uranium Oxide (UOX) and Mixed Oxide (MOX) fuels, respectively. MTL enables simultaneous learning of the evolution of the inventories for different observables – i.e., transuranium elements, fission products, minor actinides and fertile materials – offering improved predictive capabilities compared to non-MTL neural networks, as demonstrated in the cross-validation tests. Hyperparameters of the models were found using Bayesian optimization, resulting in superior model performance compared to the old CRAM-based model. Verification against SERPENT2, used as a reference code for comparison, demonstrates the stability and accuracy of the MTL-based models, outperforming the original CRAM method for the majority of predicted isotopes.
Abstract
This paper introduces an alternative approach to irradiation modelling within the context of nuclear fuel cycle codes like ANICCA, the fuel cycle code developed by the Belgian Nuclear Research Centre (SCK CEN). The focus lies on upgrading the irradiation module by substituting the CRAM-based depletion calculations for a more flexible and innovative approach based on Multi-Task Learning (MTL). Utilizing data generated with SERPENT2 Monte Carlo simulations, two MTL neural networks are developed to surrogate irradiation processes of Uranium Oxide (UOX) and Mixed Oxide (MOX) fuels, respectively. MTL enables simultaneous learning of the evolution of the inventories for different observables – i.e., transuranium elements, fission products, minor actinides and fertile materials – offering improved predictive capabilities compared to non-MTL neural networks, as demonstrated in the cross-validation tests. Hyperparameters of the models were found using Bayesian optimization, resulting in superior model performance compared to the old CRAM-based model. Verification against SERPENT2, used as a reference code for comparison, demonstrates the stability and accuracy of the MTL-based models, outperforming the original CRAM method for the majority of predicted isotopes.
참고문헌 (26)
- T. Tajima, "East meets West again in order to tackle the global energy crises", Usp. Fiz. Nauk, 192(11), pp.1280~1292 (2022)
- G. Halkos, "Energy crisis risk mitigation through nuclear power and RES as alternative solutions towards self-sufficiency", J. Risk Financ. Manag, 16(1), pp.1~ (2023)[DOI]
- I.M. Rodríguez, "The nuclear fuel cycle code ANICCA: verification and a case study for the phase out of Belgian nuclear power with minor actinide transmutation", Nucl. Eng. Technol, 52(10), pp.2274~2284 (2020)[DOI]
- A. Stankovskiy, "Advanced method for calculations of core burnup, activation of structural materials, and spallation products accumulation in accelerator-driven systems", Sci. Technol. Nucl. Install, pp.e545103~ (2012)[DOI]
- J. Leppanen, "The Serpent Monte Carlo code: status, development and applications in 2013", Ann. Nucl. Energy, 82, pp.142~150 (2015)[DOI]
- M. Pusa, "Rational approximations to the matrix exponential in burnup calculations", Nucl. Sci. Eng, 169, pp.155~167 (2011)[DOI]
- J. W. Bae, "Deep learning approach to nuclear fuel transmutation in a fuel cycle simulator", Ann. Nucl. Energy, 139, pp.107230~ (2020)[DOI]
- K.D. Huff, "Fundamental concepts in the Cyclus nuclear fuel cycle simulation framework", Adv. Eng. Software, 94, pp.46~59 (2016)[DOI]
- V.J.C. Molina, "Deep learning models as an approach to nuclear fuel irradiation processes in pressurized water reactors", 2022 41st International Conference of the Chilean Computer Science Society (SCCC), pp.1~8 (2022)
- "An overview of multi-task learning in deep neural networks" (2017)[URL]
- MOX Cross-Section Libraries for ORIGEN-ARP, "I.C. Gauld", Oak Ridge National Lab, ~ (2003)
- V.J. Casas-Molina, "Dataset of observables for UOX and MOX spent fuel extracted from Serpent2 fuel depletion calculations for PWRs", Data Brief, 49 (2023)[DOI]
- Werner von Lensa, "Schriften des Forschungszentrums Jülich Reihe Energie & Umwelt, vol. 15" (2008)
- NEA, "Benchmark Study on Nuclear Fuel Cycle Transition Scenarios Analysis Codes" (2012), OECD Publishing
- D.A. Roberts, "The Principles of Deep Learning Theory", ~ (2022)
- "Keras"[URL]
- "TensorFlow: large-scale machine learning on heterogeneous systems" (2015)[URL]
- "Deep learning using rectified linear units (ReLU)" (2019)[URL]
- "Generalized huber loss for robust learning and its efficient minimization for a robust statistics" (2021)[URL]
- J. Wu, "Hyperparameter optimization for machine learning models based on bayesian optimization", J. Electron. Sci. Technol, 17(1), pp.26~40 (2019)[DOI]
- "KerasTuner" (2019)[URL]
- "Adam: a method for stochastic optimization" (2017)[URL]
- "A survey on activation functions and their relation with xavier and He normal initialization" (2020)[URL]
- B.G. Marcot, "What is an optimal value of k in k-fold cross-validation in discrete Bayesian network analysis?", Comput. Stat, 36(3), pp.2009~2031 (2021)[DOI]
- D.M. Allen, "The relationship between variable selection and data agumentation and a method for prediction", Technometrics, 16(1), pp.125~127 (1974)[DOI]
- Operating experience with nuclear power stations in member states 2022, "International Atomic Energy Agency(IAEA)", IAEA, 42~68, pp.IAEA Annual reports 53 (2022)