簡易檢索 / 詳目顯示

回結果列表
研究生： 陳麒皓
Chen, Chi-Hao
論文名稱： 由深度學習和黏彈頻譜反算固體或液體的物理性質
Extraction of physical properties of solids and liquids through deep learning from PVS and LPVS experiment
指導教授： 王雲哲
Wang, Yun-Che
學位類別： 碩士
Master
系所名稱： 工學院 - 土木工程學系
Department of Civil Engineering
論文出版年： 2020
畢業學年度： 108
語文別： 英文
論文頁數： 89
中文關鍵詞： 鐘擺式黏彈頻譜儀 、鐘擺式黏彈流體頻譜儀 、線黏彈材料性質 、深度學習 、深度神經網路 、合成數據
外文關鍵詞： PVS, LPVS, Linear viscoelastic properties, Deep learning, Deep neural network, Synthetic dataset
相關次數： 點閱：158  下載：1 
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報

    • 鐘擺式黏彈頻譜儀(PVS)和鐘擺式黏彈流體頻譜儀(LPVS)分別可用來量測固體和液體的時間相依或頻率相依材料性質，如楊氏模數、剪力模數以及黏度。為了更精確地從實驗所得的原始資料中反算獲得材料性質，我們採用人工智慧技術中的深度神經網路(DNN)加上利用有限元素模擬及實驗產生足夠的資料集以利神經網路之訓練。我們成功地經由訓練完畢之神經網路做預測，而獲得合理且具收斂性之結果。但是在預測誤差方面仍有可進步之空間，可藉由調整深度神經網路架構及增加訓練所需之資料集來增加網路預測的準確性。本研究展示深度神經網路面對反算問題之可靠性，當古典理論解及其他反算方法失效時，深度神經網路成為一項強大的工具。本論文中，PMMA的黏彈材料性質是PVS反算問題待預測之目標，根據目前的深度神經網路所得預測誤差在楊氏模數為正負5GPa、普松比為正負0.1、正切消散模數為正負0.01。而LPVS反算問題待預測的目標則是含二氧化鋯奈米顆粒之丁酮混合溶液中的平均顆粒粒徑，根據目前的深度神經網路所得預測誤差在正負5奈米。

    PVS (Pendulum-type Viscoelastic Spectroscopy) and LPVS (Liquid Pendulum-type Viscoelastic Spectroscopy) are experimental methods to characterize, respectively, solids and liquids for their time-dependent and frequency-dependent material properties, such as Young’s
    modulus, shear modulus and viscosity. In order to better solve the inverse problems to obtain material properties from experimentally measured raw data, an artificial intelligence technique, based on deep neural networks (DNN), has been developed. Numerically calculated complex
    Young’s modulus, shear modulus and viscosity from the finite element method, in conjunction with experimental data are used to generate sufficient labeled data for DNN training. Sufficiently reasonable convergence results are obtained from the training. Improvements may be achievable through better DNN architecture design and training with more datasets. It is demonstrated that DNN methodologies are powerful to solve inverse problems when other inverse schemes are not efficient. For PVS inverse problem, properties of PMMA specimens are predicted by
    DNN, errors within 5 GPa in Young’s modulus E, 0.1 in Poisson’s ratio and 0.01 in loss tangent has been reached so far. For LPVS inverse problem, mean particle size (D50) of ZrO2/MEK suspension fluid is predicted, error within 5nm has been achieved so far.

    CHINESE ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii NOMENCLATURE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Goals and motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.3 Outline of this thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2 Theoretical foundation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2.1 Linear viscoelastic materials . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2.2 Standard linear solid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.3 Resonant frequency of a cantilever beam with end inertia . . . . . . . . . . . . 6 2.4 Loss tangent determination . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.4.1 Sub-resonant regime . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.4.2 Near resonant regime . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 3 Experimental considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 3.1 Pendulum-type Viscoelastic Spectroscopy (PVS) . . . . . . . . . . . . . . . . 10 3.1.1 Resonant regime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3.1.2 Sub-resonant regime . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 3.2 Liquid Pendulum-type Viscoelastic Spectroscopy (LPVS) . . . . . . . . . . . . 18 3.2.1 Resonant regime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.2.2 Sub-resonant regime . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.2.3 Heating experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 3.3 Fourier-transform infrared spectroscopy (FTIR) . . . . . . . . . . . . . . . . . 23 3.4 Specimen introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.4.1 PMMA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.4.2 ZrO2/MEK suspension fluid . . . . . . . . . . . . . . . . . . . . . . . 26 4 Deep learning and numerical considerations . . . . . . . . . . . . . . . . . . . . 28 4.1 Deep learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 4.2 Neural network architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 4.3 Loss function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 4.4 Stochastic gradient descent . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 4.5 Overfitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.6 PVS inverse problem by deep learning . . . . . . . . . . . . . . . . . . . . . . 35 4.6.1 PVS 3D FEM simulation . . . . . . . . . . . . . . . . . . . . . . . . . 37 4.6.2 Data post-processing . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 4.6.3 Build neural network . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 4.6.4 Training, validating and predicting . . . . . . . . . . . . . . . . . . . . 42 4.6.5 Adjust neural network or create more data . . . . . . . . . . . . . . . . 43 4.7 LPVS inverse problem by deep learning . . . . . . . . . . . . . . . . . . . . . 43 4.7.1 LPVS experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 4.7.2 Data post-processing . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 4.7.3 Build neural network . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 4.7.4 Training, validating and predicting . . . . . . . . . . . . . . . . . . . . 50 4.7.5 Adjust neural network or create more data . . . . . . . . . . . . . . . . 50 5 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 5.1 PVS inverse problem by deep learning . . . . . . . . . . . . . . . . . . . . . . 52 5.2 Testing with experimental dataset . . . . . . . . . . . . . . . . . . . . . . . . . 56 5.2.1 PMMA slender specimen . . . . . . . . . . . . . . . . . . . . . . . . . 56 5.2.2 PMMA short specimen . . . . . . . . . . . . . . . . . . . . . . . . . . 58 5.2.3 Comparison with DNN prediction . . . . . . . . . . . . . . . . . . . . 59 5.3 Summary of PVS inverse problem . . . . . . . . . . . . . . . . . . . . . . . . 62 5.4 LPVS inverse problem by deep learning . . . . . . . . . . . . . . . . . . . . . 64 5.5 Summary of LPVS inverse problem . . . . . . . . . . . . . . . . . . . . . . . 68 6 Conclusion and future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 6.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 6.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 LIST OF REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 APPENDICES Appendix A: Python code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 VITA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

    [1] R. S. Lakes. Viscoelastic materials. Cambridge University Press, 2009.
    [2] H. Bi, S. Meng, Y. Li, K. Guo, Y. Chen, J. Kong, P. Yang, W. Zhong, and B. Liu. Deposition
    of peg onto pmma microchannel surface to minimize nonspecific adsorption. Lab on a chip, 6:769–75, 07 2006.
    [3] T. C. Kao. Particle size analysis of nanofluids via liquid pendulum type viscoelastic spectroscopy. Master’s thesis, National Cheng Kung University civil engineering, 2018.
    [4] K. R. Yang. Study of pendulum-type viscoelastic spectroscopy for solids and liquids. Master’s thesis, National Cheng Kung University civil engineering, 2019.
    [5] F. Chollet. Deep learning with Python. Manning Publications, 2018.
    [6] J. Woirgard, Y. Sarrazin, and H. Chaumet. Apparatus for the measurement of internal friction as a function of frequency between 10-5 and 10 Hz. Review of Scientific Instruments, 48(10):1322, 1977.
    [7] M. Brodt, L. S. Cook, and R. S. Lakes. Apparatus for measuring viscoelastic properties over ten decades: refinements. Journal of Materials Science, 66(11):5292–5297, 1995.
    [8] Y. C. Wang, C. C. Ko, H. K. Wu, and Y. D. Wu. Pendulum-type viscoelastic spectroscopy for damping measurement of solids. Japan Society of Mechanical Engineers, 13:s137–s142, 2013.
    [9] Y. C. Wang, C. C. Ko, and L. M. Shiau. Accurate determination of torsion and pure bending moment for viscoelastic measurements. International Journal of Modern Physics: Conference Series, 24:1360016, 2013.
    [10] P. Kao. Studies of pendulum-type viscoelastic spectroscopy for fluids. Master’s thesis, National Cheng Kung University civil engineering, 2017.
    [11] M. T. Hagan, H. B. Demuth, M. H. Beale, and O. D. Jesus. Neural network design. Martin Hagan, 2 edition, 2014.
    [12] Mart´ın Abadi, Ashish Agarwal, Paul Barham, Eugene Brevdo, Zhifeng Chen, Craig Citro, Greg S. Corrado, Andy Davis, Jeffrey Dean, Matthieu Devin, Sanjay Ghemawat, Ian
    Goodfellow, Andrew Harp, Geoffrey Irving, Michael Isard, Yangqing Jia, Rafal Jozefowicz, Lukasz Kaiser, Manjunath Kudlur, Josh Levenberg, Dan Man´e, Rajat Monga, Sherry
    Moore, Derek Murray, Chris Olah, Mike Schuster, Jonathon Shlens, Benoit Steiner, Ilya Sutskever, Kunal Talwar, Paul Tucker, Vincent Vanhoucke, Vijay Vasudevan, Fernanda
    Vi´egas, Oriol Vinyals, Pete Warden, Martin Wattenberg, Martin Wicke, Yuan Yu, and Xiaoqiang Zheng. TensorFlow: Large-scale machine learning on heterogeneous systems,
    2015. Software available from tensorflow.org.
    [13] I. Kopal, I. Labaj, M. Harnicarova, J. Valicek, and D. Hruby. Prediction of the tensile response of carbon black filled rubber blends by artificial neural network. Polymers, 10:644, 06 2018.
    [14] I. Kopal, M. Harnicarova, J. Valicek, J. Krmela, and O. Lukac. Radial basis function
    neural network-based modeling of the dynamic thermo-mechanical response and damping behavior of thermoplastic elastomer systems. Polymers, 11:1074, 06 2019.
    [15] L. Lu, M. Dao, P. Kumar, U. Ramamurty, G. E. Karniadakis, and S. Suresh. Extraction of mechanical properties of materials through deep learning from instrumented indentation. Proceedings of the National Academy of Sciences, 117(13):7052–7062, 2020.

    下載圖示 校內：2025-07-01公開
    校外：2025-07-01公開
    QR CODE