隨著現今科技對於多功能結構材料的需求日漸增加，且伴隨著全球暖化使環保意識崛起，如何使結構材料擁有多功能性質且能減少材料消耗是目前許多工程領域的重要研究方向。在自然界中，生物材料能藉由不同材料間的排列或幾何結構組成又或是以軟硬材料組成的複合材料來表現出優異的材料性質，例如於蝴蝶翅膀的組成中發現的Gyroid結構，Gyroid是一種由連續的曲面組成且符合高能量吸收、高強度等性質的孔洞結構，能藉由更改壁厚決定所具備的材料性質。仿生材料的複雜結構會造成模擬或結構設計上的不易使得計算量增加，而透過人工智慧技術可以有效率地進行設計。深度學習在經過訓練後可以從過去的經驗中學習如何決策或預測，發揮強大的運算力計算龐大資訊量，將複雜系統簡化，強化學習則可以透過在與環境迭代過程中藉由性質的變化學習設計的策略，能夠更穩定且有效的給出最佳化設計。
本研究建立了一套設計結構材料的方法，以仿生材料為基礎設計具備高耐衝擊特性且同時又具備輕量化的材料。為了解決仿生材料在計算上會耗費大量的運算時間，本研究利用簡化模型取代原先的複雜結構使其具備同樣的結構性質，以解決有限元模擬中的複雜幾何問題，並應用深度強化學習模型加入多目標設計，可以依照需求指定相對應的性質對軟硬材料分布進行最佳化設計。其中，將強化學習與有限元模擬相互結合，可以在迭代過程判斷設計的性質，以大幅的減少設計成本。此設計方法可以讓設計者能夠根據特定需求進行複合材料設計，同時強化學習與有限元模擬也能根據需求更換設計空間與最佳化目標。因此，本研究所提出的設計方法具有高度的靈活性，並能適用於不同的孔洞結構，解決複雜結構材料難以進行設計的問題。

In nature, biological materials demonstrate exceptional performance through different arrangements of materials or geometric structures, as seen in the Gyroid structure found in butterfly wings. Composed of continuous surfaces, the Gyroid exhibits properties like high energy absorption and strength, with its material properties determined by the thickness of its structure. However, the complexity of bioinspired materials presents challenges in simulation and structural design, leading to increased computational costs. To address this, artificial intelligence (AI) technologies can be utilized.
This study establishes a method for designing structural materials based on bioinspired principles, aiming to create materials with high-impact resistance while maintaining lightweight properties. Integrating deep reinforcement learning models with multi-objective design empowers designers to specify desired material properties and optimize the distribution of soft and stiff materials according to specific requirements. The combination of reinforcement learning and finite element simulation allows the assessment of design properties, resulting in significant reductions in computational costs.
This design method enables designers to customize composite materials to meet specific needs, as both reinforcement learning and finite element simulation can adapt to changing design spaces and optimization objectives. As a result, the proposed method exhibits high flexibility, applicable to various porous structures, effectively addressing the challenges of designing complex structural materials.

[1] U. G. K. Wegst, H. Bai, E. Saiz, A. P. Tomsia, and R. O. Ritchie, ‘Bioinspired structural materials’, Nat Mater, vol. 14, no. 1, pp. 23–36, 2015.
[2] K. Liu and L. Jiang, ‘Bio-inspired design of multiscale structures for function integration’, Nano Today, vol. 6, no. 2, pp. 155–175, 2011.
[3] N. K. Katiyar, G. Goel, S. Hawi, and S. Goel, ‘Nature-inspired materials: Emerging trends and prospects’, NPG Asia Mater, vol. 13, no. 1, p. 56, 2021.
[4] V. Saranathan et al., ‘Structural diversity of arthropod biophotonic nanostructures spans amphiphilic phase-space’, Nano Lett, vol. 15, no. 6, pp. 3735–3742, 2015.
[5] Y. Tian et al., ‘Sea urchin skeleton-inspired triply periodic foams for fast latent heat storage’, Int J Heat Mass Transf, vol. 206, p. 123944, 2023.
[6] Y. Kobayashi, R. Ohnuki, and S. Yoshioka, ‘Discovery of I-WP minimal-surface-based photonic crystal in the scale of a longhorn beetle’, J R Soc Interface, vol. 18, no. 184, p. 20210505, 2021.
[7] K. Michielsen and D. G. Stavenga, ‘Gyroid cuticular structures in butterfly wing scales: biological photonic crystals’, J R Soc Interface, vol. 5, no. 18, pp. 85–94, 2008.
[8] C. Mille, E. C. Tyrode, and R. W. Corkery, ‘Inorganic chiral 3-D photonic crystals with bicontinuous gyroid structure replicated from butterfly wing scales’, Chemical Communications, vol. 47, no. 35, pp. 9873–9875, 2011.
[9] C. Pouya and P. Vukusic, ‘Electromagnetic characterization of millimetre-scale replicas of the gyroid photonic crystal found in the butterfly Parides sesostris’, Interface Focus, vol. 2, no. 5, pp. 645–650, 2012.
[10] J. A. Dolan, B. D. Wilts, S. Vignolini, J. J. Baumberg, U. Steiner, and T. D. Wilkinson, ‘Optical properties of gyroid structured materials: from photonic crystals to metamaterials’, Adv Opt Mater, vol. 3, no. 1, pp. 12–32, 2015.
[11] A. Yánez, A. Herrera, O. Martel, D. Monopoli, and H. Afonso, ‘Compressive behaviour of gyroid lattice structures for human cancellous bone implant applications’, Materials Science and Engineering: C, vol. 68, pp. 445–448, 2016.
[12] I. Maskery, N. T. Aboulkhair, A. O. Aremu, C. J. Tuck, and I. A. Ashcroft, ‘Compressive failure modes and energy absorption in additively manufactured double gyroid lattices’, Addit Manuf, vol. 16, pp. 24–29, 2017.
[13] M. Saba, B. D. Wilts, J. Hielscher, and G. E. Schröder-Turk, ‘Absence of circular polarisation in reflections of butterfly wing scales with chiral gyroid structure’, Mater Today Proc, vol. 1, pp. 193–208, 2014.
[14] S. L. Brunton, M. Budišić, E. Kaiser, and J. N. Kutz, ‘Modern Koopman theory for dynamical systems’, arXiv preprint arXiv:2102.12086, 2021.
[15] S. Lu, Q. Zhou, Y. Ouyang, Y. Guo, Q. Li, and J. Wang, ‘Accelerated discovery of stable lead-free hybrid organic-inorganic perovskites via machine learning’, Nat Commun, vol. 9, no. 1, p. 3405, 2018.
[16] K. T. Butler, D. W. Davies, H. Cartwright, O. Isayev, and A. Walsh, ‘Machine learning for molecular and materials science’, Nature, vol. 559, no. 7715, pp. 547–555, 2018.
[17] P. Z. Moghadam et al., ‘Structure-mechanical stability relations of metal-organic frameworks via machine learning’, Matter, vol. 1, no. 1, pp. 219–234, 2019.
[18] C. Park, C. C. Took, and J.-K. Seong, ‘Machine learning in biomedical engineering’, Biomed Eng Lett, vol. 8, pp. 1–3, 2018.
[19] Z. Yang, C.-H. Yu, and M. J. Buehler, ‘Deep learning model to predict complex stress and strain fields in hierarchical composites’, Sci Adv, vol. 7, no. 15, p. eabd7416, 2021.
[20] Z. Yang, C.-H. Yu, K. Guo, and M. J. Buehler, ‘End-to-end deep learning method to predict complete strain and stress tensors for complex hierarchical composite microstructures’, J Mech Phys Solids, vol. 154, p. 104506, 2021.
[21] D. Amodei et al., ‘Deep speech 2: End-to-end speech recognition in english and mandarin’, in International conference on machine learning, PMLR, 2016, pp. 173–182.
[22] A. B. Nassif, I. Shahin, I. Attili, M. Azzeh, and K. Shaalan, ‘Speech recognition using deep neural networks: A systematic review’, IEEE access, vol. 7, pp. 19143–19165, 2019.
[23] L. Deng and X. Li, ‘Machine learning paradigms for speech recognition: An overview’, IEEE Trans Audio Speech Lang Process, vol. 21, no. 5, pp. 1060–1089, 2013.
[24] C.-H. Yu, Z. Qin, F. J. Martin-Martinez, and M. J. Buehler, ‘A self-consistent sonification method to translate amino acid sequences into musical compositions and application in protein design using artificial intelligence’, ACS Nano, vol. 13, no. 7, pp. 7471–7482, 2019.
[25] C.-H. Yu, E. Khare, O. P. Narayan, R. Parker, D. L. Kaplan, and M. J. Buehler, ‘ColGen: An end-to-end deep learning model to predict thermal stability of de novo collagen sequences’, J Mech Behav Biomed Mater, vol. 125, p. 104921, 2022.
[26] S. L. Franjou, M. Milazzo, C.-H. Yu, and M. J. Buehler, ‘Sounds interesting: Can sonification help us design new proteins?’, Expert Rev Proteomics, vol. 16, no. 11–12, pp. 875–879, 2019.
[27] Z. Yang, C.-H. Yu, and M. J. Buehler, ‘Deep learning model to predict complex stress and strain fields in hierarchical composites’, Sci Adv, vol. 7, no. 15, p. eabd7416, 2021.
[28] C.-H. Yu et al., ‘End-to-end deep learning model to predict and design secondary structure content of structural proteins’, ACS Biomater Sci Eng, vol. 8, no. 3, pp. 1156–1165, 2022.
[29] C. Yu et al., ‘Hierarchical Multiresolution Design of Bioinspired Structural Composites Using Progressive Reinforcement Learning’, Adv Theory Simul, vol. 5, no. 11, p. 2200459, 2022.
[30] D. Dewey, ‘Reinforcement learning and the reward engineering principle’, in 2014 AAAI Spring Symposium Series, 2014.
[31] P. Dayan and B. W. Balleine, ‘Reward, motivation, and reinforcement learning’, Neuron, vol. 36, no. 2, pp. 285–298, 2002.
[32] M. Mahmud, M. S. Kaiser, A. Hussain, and S. Vassanelli, ‘Applications of deep learning and reinforcement learning to biological data’, IEEE Trans Neural Netw Learn Syst, vol. 29, no. 6, pp. 2063–2079, 2018.
[33] L. P. Kaelbling, M. L. Littman, and A. W. Moore, ‘Reinforcement learning: A survey’, Journal of artificial intelligence research, vol. 4, pp. 237–285, 1996.
[34] Y. Li, ‘Deep reinforcement learning: An overview’, arXiv preprint arXiv:1701.07274, 2017.
[35] Z. Gan, M. D. Turner, and M. Gu, ‘Biomimetic gyroid nanostructures exceeding their natural origins’, Sci Adv, vol. 2, no. 5, p. e1600084, 2016.
[36] D. W. Abueidda, M. Elhebeary, C.-S. A. Shiang, S. Pang, R. K. A. Al-Rub, and I. M. Jasiuk, ‘Mechanical properties of 3D printed polymeric Gyroid cellular structures: Experimental and finite element study’, Mater Des, vol. 165, p. 107597, 2019.
[37] S. Hyde et al., The language of shape: the role of curvature in condensed matter: physics, chemistry and biology. Elsevier, 1996.
[38] L. Han and S. Che, ‘An overview of materials with triply periodic minimal surfaces and related geometry: From biological structures to self‐assembled systems’, Advanced Materials, vol. 30, no. 17, p. 1705708, 2018.
[39] L. Schläfli, Gesammelte Mathematische Abhandlungen, vol. 1. Springer, 1950.
[40] A. H. Schoen, ‘Infinite periodic minimal surfaces without self-intersections’, 1970.
[41] J. W. Galusha, L. R. Richey, J. S. Gardner, J. N. Cha, and M. H. Bartl, ‘Discovery of a diamond-based photonic crystal structure in beetle scales’, Phys Rev E, vol. 77, no. 5, p. 050904, 2008.
[42] G. E. Schröder-Turk et al., ‘The chiral structure of porous chitin within the wing-scales of Callophrys rubi’, J Struct Biol, vol. 174, no. 2, pp. 290–295, 2011.
[43] L. J. Gibson, ‘Cellular solids’, MRS Bull, vol. 28, no. 4, pp. 270–274, 2003.
[44] M. F. Ashby and L. J. Gibson, ‘Cellular solids: structure and properties’, Press Syndicate of the University of Cambridge, Cambridge, UK, pp. 175–231, 1997.
[45] J. Chakrabarty, Theory of plasticity. Elsevier, 2012.
[46] T. Tajmajer, ‘Modular multi-objective deep reinforcement learning with decision values’, in 2018 Federated conference on computer science and information systems (FedCSIS), IEEE, 2018, pp. 85–93.