UNCERTAINTY QUANTIFICATION OF DEEP NEURAL NETWORK-BASED TURBULENCE MODEL FOR REACTOR TRANSIENT ANALYSIS

被引:0
|
作者
Liu, Yang [1 ]
Hu, Rui [1 ]
Balaprakash, Prasanna [2 ]
机构
[1] Argonne Natl Lab, Nucl Sci & Engn Div, Lemont, IL 60439 USA
[2] Argonne Natl Lab, Math & Comp Sci Div, Lemont, IL USA
来源
PROCEEDINGS OF THE 2021 ASME VERIFICATION AND VALIDATION SYMPOSIUM (VVS2021) | 2021年
关键词
deep learning; turbulence model; Monte Carlo dropout; deep ensemble; Bayesian neural network;
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Deep neural networks (DNNs) have demonstrated good performance in learning highly non-linear relationships in large datasets, thus have been considered as a promising surrogate modeling tool for parametric partial differential equations (PDEs). On the other hand, quantifying the predictive uncertainty in DNNs is still a challenging problem. The Bayesian neural network (BNN), a sophisticated method assuming the weights of the DNNs follow certain uncertainty distributions, is considered as a state-of-the-art method for the UQ of DNNs. However, the method is too computationally expensive to be used in complicated DNN architectures. In this work, we utilized two more methods for the UQ of complicated DNNs, i.e. Monte Carlo dropout and deep ensemble. Both methods are computationally efficient and scalable compared to BNN. We applied these two methods to a densely connected convolutional network, which is developed and trained as a coarse-mesh turbulence closure relation for reactor safety analysis. In comparison, the corresponding BNN with the same architecture is also developed and trained. The computational cost and uncertainty evaluation performance of these three UQ methods are comprehensively investigated. It is found that the deep ensemble method is able to produce reasonable uncertainty estimates with good scalability and relatively low computational cost compared to BNN.
引用
收藏
页数:7
相关论文
共 50 条
  • [1] Uncertainty quantification of deep neural network-based turbulence model for reactor transient analysis
    Liu, Yang
    Hu, Rui
    Balaprakash, Prasanna
    Proceedings of the 2021 ASME Verification and Validation Symposium, VVS 2021, 2021,
  • [2] Partial Adversarial Training for Neural Network-Based Uncertainty Quantification
    Kabir, H. M. Dipu
    Khosravi, Abbas
    Nahavandi, Saeid
    Kavousi-Fard, Abdollah
    IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTATIONAL INTELLIGENCE, 2021, 5 (04): : 595 - 606
  • [3] Neural Network-Based Uncertainty Quantification: A Survey of Methodologies and Applications
    Kabir, H. M. Dipu
    Khosravi, Abbas
    Hosen, Mohammad Anwar
    Nahavandi, Saeid
    IEEE ACCESS, 2018, 6 : 36218 - 36234
  • [4] Deep neural network-based classification model for Sentiment Analysis
    Pan, Donghang
    Yuan, Jingling
    Li, Lin
    Sheng, Deming
    2019 6TH INTERNATIONAL CONFERENCE ON BEHAVIORAL, ECONOMIC AND SOCIO-CULTURAL COMPUTING (BESC 2019), 2019,
  • [5] Neural Network-Based Metamodel of synthetic seismograms: Application for uncertainty quantification
    Noureddine, Mohamad Ali
    De Martin, Florent
    El Meouche, Rani
    Sammuneh, Muhammad Ali
    Ababsa, Fakhreddine
    Beaufils, Mickael
    ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE, 2025, 151
  • [6] Breast cancer classification by a new approach to assessing deep neural network-based uncertainty quantification methods
    Hamedani-KarAzmoudehFar, Fatemeh
    Tavakkoli-Moghaddam, Reza
    Tajally, Amir Reza
    Aria, Seyed Sina
    BIOMEDICAL SIGNAL PROCESSING AND CONTROL, 2023, 79
  • [7] Global Surrogate Modeling by Neural Network-Based Model Uncertainty
    Leifsson, Leifur
    Nagawkar, Jethro
    Barnet, Laurel
    Bryden, Kenneth
    Koziel, Slawomir
    Pietrenko-Dabrowska, Anna
    COMPUTATIONAL SCIENCE - ICCS 2022, PT III, 2022, 13352 : 425 - 434
  • [8] Deep Convolution Neural Network-Based Crack Feature Extraction, Detection and Quantification
    Shuai Teng
    Gongfa Chen
    Journal of Failure Analysis and Prevention, 2022, 22 : 1308 - 1321
  • [9] Deep Convolution Neural Network-Based Crack Feature Extraction, Detection and Quantification
    Teng, Shuai
    Chen, Gongfa
    JOURNAL OF FAILURE ANALYSIS AND PREVENTION, 2022, 22 (03) : 1308 - 1321
  • [10] DBoTPM: A Deep Neural Network-Based Botnet Prediction Model
    Haq, Mohd Anul
    ELECTRONICS, 2023, 12 (05)