A non-intrusive approach for physics-constrained learning with application to fuel cell modeling

被引:0
|
作者
Vishal Srivastava
Valentin Sulzer
Peyman Mohtat
Jason B. Siegel
Karthik Duraisamy
机构
[1] University of Michigan,Aerospace Engineering
[2] University of Michigan,Mechanical Engineering
来源
Computational Mechanics | 2023年 / 72卷
关键词
Data-driven modeling; Machine learning; Model augmentation; Statistical inference;
D O I
暂无
中图分类号
学科分类号
摘要
A data-driven model augmentation framework, referred to as weakly-coupled Integrated Inference and Machine Learning (IIML), is presented to improve the predictive accuracy of physical models. In contrast to parameter calibration, this work seeks corrections to the structure of the model by (a) inferring augmentation fields that are consistent with the underlying model, and (b) transforming these fields into corrective model forms. The proposed approach couples the inference and learning steps in a weak sense via an alternating optimization approach. This coupling ensures that the augmentation fields remain learnable and maintain consistent functional relationships with local modeled quantities across the training dataset. An iterative solution procedure is presented in this paper, removing the need to embed the augmentation function during the inference process. This framework is used to infer an augmentation introduced within a polymer electrolyte membrane fuel cell (PEMFC) model using a small amount of training data (from only 14 training cases). These training cases belong to a dataset consisting of high-fidelity simulation data obtained from a high-fidelity model of a first generation Toyota Mirai. All cases in this dataset are characterized by different inflow and outflow conditions on the same geometry. When tested on 1224 different configurations, the inferred augmentation significantly improves the predictive accuracy for a wide range of physical conditions. Predictions and available data for the current density distribution are also compared to demonstrate the predictive capability of the model for quantities of interest which were not involved in the inference process. The results demonstrate that the weakly-coupled IIML framework offers sophisticated and robust model augmentation capabilities without requiring extensive changes to the numerical solver.
引用
收藏
页码:411 / 430
页数:19
相关论文
共 50 条
  • [41] Transfer Learning for Non-Intrusive Load Monitoring
    D'Incecco, Michele
    Squartini, Stefano
    Zhong, Mingjun
    IEEE TRANSACTIONS ON SMART GRID, 2020, 11 (02) : 1419 - 1429
  • [42] A Non-intrusive Approach for Smart Power Meter
    Tavares, Hugo
    Prado, Bruno
    Bispo, Kalil
    Dantas, Daniel
    2018 IEEE 16TH INTERNATIONAL CONFERENCE ON INDUSTRIAL INFORMATICS (INDIN), 2018, : 605 - 610
  • [43] A non-intrusive isolation approach for soft cores
    Sinanoglu, Ozgur
    Petrov, Tsvetomir
    2007 DESIGN, AUTOMATION & TEST IN EUROPE CONFERENCE & EXHIBITION, VOLS 1-3, 2007, : 27 - +
  • [44] Towards the Fusion of Intrusive and Non-intrusive Load Monitoring - A Hybrid Approach
    Voelker, Benjamin
    Scholl, Philipp M.
    Schubert, Tobias
    Becker, Bernd
    E-ENERGY'18: PROCEEDINGS OF THE 9TH ACM INTERNATIONAL CONFERENCE ON FUTURE ENERGY SYSTEMS, 2018, : 436 - 438
  • [45] A Semi-supervised and Online Learning Approach for Non-Intrusive Load Monitoring
    Salem, Hajer
    Sayed-Mouchaweh, Moamar
    MACHINE LEARNING AND KNOWLEDGE DISCOVERY IN DATABASES, ECML PKDD 2019, PT III, 2020, 11908 : 585 - 601
  • [46] Recognition and classification of typical load profiles in buildings with non-intrusive learning approach
    Piscitelli, Marco Savino
    Brandi, Silvio
    Capozzoli, Alfonso
    APPLIED ENERGY, 2019, 255
  • [47] Physics-constrained Data-Driven Variational method for discrepancy modeling
    Masud, Arif
    Nashar, Sharbel
    Goraya, Shoaib A.
    COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING, 2023, 417
  • [48] A physics-constrained and data-driven method for modeling supersonic flow
    Zhao, Tong
    Xu, Yuming
    He, Guoqiang
    Qin, Fei
    PHYSICS OF FLUIDS, 2024, 36 (06)
  • [49] Physics-constrained Automatic Feature Engineering for Predictive Modeling in Materials Science
    Xiang, Ziyu
    Fan, Mingzhou
    Tovar, Guillermo Vazquez
    Trehem, William
    Yoon, Byung-Jun
    Qian, Xiaofeng
    Arroyave, Raymundo
    Qian, Xiaoning
    THIRTY-FIFTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THIRTY-THIRD CONFERENCE ON INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE AND THE ELEVENTH SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2021, 35 : 10414 - 10421
  • [50] Physics-Constrained Machine Learning for Reliability-Based Design Optimization
    Xu, Yanwen
    Wang, Pingfeng
    2023 ANNUAL RELIABILITY AND MAINTAINABILITY SYMPOSIUM, RAMS, 2023,