Automated and efficient local adaptive regression for principal component-based reduced-order modeling of turbulent reacting flows

被引:6
作者
D'Alessio, Giuseppe [1 ]
Sundaresan, Sankaran [2 ]
Mueller, Michael E. [1 ]
机构
[1] Princeton Univ, Dept Mech & Aerosp Engn, Princeton, NJ 08544 USA
[2] Princeton Univ, Dept Chem & Biol Engn, Princeton, NJ 08544 USA
关键词
Machine learning; Principal Component Analysis; Local regression; Bayesian optimization; Turbulent flames; COMBUSTION; FLAMES; SOOT;
D O I
10.1016/j.proci.2022.07.235
中图分类号
O414.1 [热力学];
学科分类号
摘要
Principal Component Analysis can be used to reduce the cost of Computational Fluid Dynamics simulations of turbulent reacting flows by reducing the dimensionality of the transported variables through projection of the thermochemical state onto a lower-dimensional manifold. However, because of the nonlinearity of the principal component source terms, nonlinear regression techniques must be utilized for the source terms in terms of the principal components. Unfortunately, widely available and utilized nonlinear regression techniques can have prohibitive computational requirements and/or accuracy that is highly dependent on user experience in ad hoc tuning of model architecture and hyperparameters. In this work, a new nonlinear regression approach is proposed that is both computationally efficient and automated so does not require any user input. The approach is evaluated through a priori prediction of principal component source terms using data from a Direct Numerical Simulation of a turbulent nonpremixed n -heptane/air jet flame. In particular, the proposed framework consists of local regressions whose complexity is adapted according to the local nonlinearity of the data: local linear regression when accurate enough and local Artificial Neural Networks when nonlinear regression is required. The number of local clusters for local regression is determined automatically using the Davies-Bouldin index. In addition, Bayesian optimization is utilized for model training (i.e., to select the best architectures and hyperparameters of the nonlinear regressions in an unsupervised fashion), eliminating ad hoc hand-tuning and/or expensive grid searches. Overall, compared to a single, global neural network, the new local adaptive regression approach is shown to have comparable accuracy but 69% less training time due to the utilization of local linear regression and faster training of local neural networks. & COPY; 2022 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
引用
收藏
页码:5249 / 5258
页数:10
相关论文
共 39 条
  • [1] Formation, growth, and transport of soot in a three-dimensional turbulent non-premixed jet flame
    Attili, Antonio
    Bisetti, Fabrizio
    Mueller, Michael E.
    Pitsch, Heinz
    [J]. COMBUSTION AND FLAME, 2014, 161 (07) : 1849 - 1865
  • [2] Data-driven Classification and Modeling of Combustion Regimes in Detonation Waves
    Barwey, Shivam
    Prakash, Supraj
    Hassanaly, Malik
    Raman, Venkat
    [J]. FLOW TURBULENCE AND COMBUSTION, 2021, 106 (04) : 1065 - 1089
  • [3] On the formation and early evolution of soot in turbulent nonpremixed flames
    Bisetti, Fabrizio
    Blanquart, Guillaume
    Mueller, Michael E.
    Pitsch, Heinz
    [J]. COMBUSTION AND FLAME, 2012, 159 (01) : 317 - 335
  • [4] Caruana R, 2001, ADV NEUR IN, V13, P402
  • [5] Application of machine learning for filtered density function closure in MILD combustion
    Chen, Zhi X.
    Iavarone, Salvatore
    Ghiasi, Golnoush
    Kannan, Veeraraghavan
    D'Alessio, Giuseppe
    Parente, Alessandro
    Swaminathan, Nedunchezhian
    [J]. COMBUSTION AND FLAME, 2021, 225 : 160 - 179
  • [6] Basic Enhancement Strategies When Using Bayesian Optimization for Hyperparameter Tuning of Deep Neural Networks
    Cho, Hyunghun
    Kim, Yongjin
    Lee, Eunjung
    Choi, Daeyoung
    Lee, Yongjae
    Rhee, Wonjong
    [J]. IEEE ACCESS, 2020, 8 : 52588 - 52608
  • [7] Numerical Modeling of Laminar Flames with Detailed Kinetics Based on the Operator-Splitting Method
    Cuoci, Alberto
    Frassoldati, Alessio
    Faravelli, Tiziano
    Ranzi, Eliseo
    [J]. ENERGY & FUELS, 2013, 27 (12) : 7730 - 7753
  • [8] Feature extraction and artificial neural networks for the on-the-fly classification of high-dimensional thermochemical spaces in adaptive-chemistry simulations
    D'Alessio, Giuseppe
    Cuoci, Alberto
    Parente, Alessandro
    [J]. DATA-CENTRIC ENGINEERING, 2021, 2
  • [9] Impact of the Partitioning Method on Multidimensional Adaptive-Chemistry Simulations
    D'Alessio, Giuseppe
    Cuoci, Alberto
    Aversano, Gianmarco
    Bracconi, Mauro
    Stagni, Alessandro
    Parente, Alessandro
    [J]. ENERGIES, 2020, 13 (10)
  • [10] Adaptive chemistry via pre-partitioning of composition space and mechanism reduction
    D'Alessio, Giuseppe
    Parente, Alessandro
    Stagni, Alessandro
    Cuoci, Alberto
    [J]. COMBUSTION AND FLAME, 2020, 211 (211) : 68 - 82