When physics-informed data analytics outperforms black-box machine learning: A case study in thickness control for additive manufacturing

被引:10
作者
Wang, Ke [1 ]
Zeng, Minxiang [2 ,3 ]
Wang, Jialu [1 ]
Shang, Wenjie [2 ]
Zhang, Yanliang [2 ]
Luo, Tengfei [2 ]
Dowling, Alexander W. [1 ]
机构
[1] Univ Notre Dame, Dept Chem & Biomol Engn, Notre Dame, IN 46556 USA
[2] Univ Notre Dame, Dept Aerosp & Mech Engn, Notre Dame, IN 46556 USA
[3] Texas Tech Univ, Dept Chem Engn, Lubbock, TX 79409 USA
来源
DIGITAL CHEMICAL ENGINEERING | 2023年 / 6卷
基金
美国国家科学基金会;
关键词
Aerosol jet printing; Model-based design of experiment; Data science; Uncertainty quantification; Gaussian process regression; DESIGN;
D O I
10.1016/j.dche.2022.100076
中图分类号
TQ [化学工业];
学科分类号
0817 ;
摘要
Aerosol jet printing (AJP) has emerged as a promising noncontact additive manufacturing method for high-resolution printing for a wide range of material systems. A key challenge limiting the broader adoption of AJP in the material science community is the lack of methods to precisely control thickness. Herein, we develop a model-based design of experiment (MBDoE) framework that integrates physics-informed models, nonlinear regression, and information criteria to postulate, select and calibrate the best model to describe and optimize the AJP manufacturing process. Starting with already available data from system commissioning (e.g., prior single variable sensitivity analysis), four candidate physics-informed models are postulated and trained. MBDoE identifies a single additional optimal experiment to validate these predictive models with quantified uncertainties, which are then used to determine the best experimental conditions to control printed film thickness. As a comparative benchmark, the analysis is repeated using the same dataset with nonparametric Gaussian process regression (GPR) model that does not incorporate physical information. Using MBDoE principles, we find that only five experiments are necessary to calibrate the nonlinear physics-informed parametric model, and with said limited data, this model outperforms the black-box machine learning GPR model. This key result underscores an emerging trend in the data science community: incorporating physical information into predictive models often drastically reduces the data requirements. Leveraging MBDoE further increased the data efficiency. By design, the proposed data science framework is general in nature and can be easily extended to other experimental and additive manufacturing systems beyond AJP.
引用
收藏
页数:17
相关论文
共 50 条
  • [1] Allison P.D., 1999, MULTIPLE REGRESSION
  • [2] Statistical tools for optimal dynamic model building
    Asprey, SP
    Macchietto, S
    [J]. COMPUTERS & CHEMICAL ENGINEERING, 2000, 24 (2-7) : 1261 - 1267
  • [3] Bard Y., 1974, Nonlinear parameter estimation
  • [4] Analytical Investigation of Aerosol Jet Printing
    Binder, Sebastian
    Glatthaar, Markus
    Raedlein, Edda
    [J]. AEROSOL SCIENCE AND TECHNOLOGY, 2014, 48 (09) : 924 - 929
  • [5] Bishop C. M., 2007, Pattern Recognition and Machine Learning Information Science and Statistics, V1st
  • [6] Managing a portfolio of interdependent new product candidates in the pharmaceutical industry
    Blau, GE
    Pekny, JF
    Varma, VA
    Bunch, PR
    [J]. JOURNAL OF PRODUCT INNOVATION MANAGEMENT, 2004, 21 (04) : 227 - 245
  • [7] On the design of optimally informative dynamic experiments for model discrimination in multiresponse nonlinear situations
    Chen, BH
    Asprey, SP
    [J]. INDUSTRIAL & ENGINEERING CHEMISTRY RESEARCH, 2003, 42 (07) : 1379 - 1390
  • [8] On the relationship between sloppiness and identifiability
    Chis, Oana-Teodora
    Villaverde, Alejandro F.
    Banga, Julio R.
    Balsa-Canto, Eva
    [J]. MATHEMATICAL BIOSCIENCES, 2016, 282 : 147 - 161
  • [9] Hybrid printing of wearable piezoelectric sensors
    Du, Yipu
    Wang, Ruoxing
    Zeng, Minxiang
    Xu, Shujia
    Saeidi-Javash, Mortaza
    Wu, Wenzhuo
    Zhang, Yanliang
    [J]. NANO ENERGY, 2021, 90
  • [10] Eugene EA, 2020, P AMER CONTR CONF, P3997, DOI [10.23919/ACC45564.2020.9148007, 10.23919/acc45564.2020.9148007]