2019-08-012024-05-18https://scholars.lib.ntu.edu.tw/handle/123456789/700157摘要：數據驅動計算力學顛覆傳統的計算力學架構，賦予計算力學在材料大數據時代的重大轉變。傳統計算力學的架構透過實驗量測材料數據，經由建立材料模型預測結構行為。在此架構下，材料模型通常是不確定性與誤差的來源，也造成材料數據與計算力學的斷鏈。數據驅動計算力學直接使用材料的應力應變數據求解，不再使用材料組成律。透過此嶄新架構，材料數據直接與泛用的力學定律結合求解，一舉解決過往依賴材料模型造成的困境。本研究的目的即是發展與建構數據驅動計算力學所需的開發平台，並將其應用於複雜的金屬塑性力學之模擬與分析。本計劃預期以三年進行，第一年將建立數據驅動計算力學的開發平台，實現數據驅動的計算核心，並以影像量測技術為基礎，發展材料數據鑑定方法，建立所需的材料數據。第二年與第三年將以複雜的金屬塑性力學之模擬與分析為標的，第二年將著重在巨觀連體尺度之金屬塑性計算力學研究，將數據驅動計算核心擴展到暫態非彈性。第三年將探討多尺度材料模擬的連結，以微尺度的晶體塑性模擬生成材料數據，並直接連結跨尺度計算，分析複雜的金屬塑性行為。數據驅動計算力學的發展仍在萌芽期，許多研究議題都仍在發展與辯證，有待進一步突破。我們希望透過此三年的努力，能在此嶄新領域做出國際領先的貢獻。<br> Abstract: Data-driven computational mechanics revolutionizes computational mechanicsframework, giving computational mechanics a major shift in the era of materialsbig data. The framework of traditional computational mechanics predictsstructural behavior using materials constitutive models fitted from materials data.Under this framework, the materials constitutive model is usually the majorsource of uncertainty and error, and also causes the broken pipe betweenmaterials data and computational mechanics. The data-driven computationalmechanics directly solves the problems using stress-strain materials data, andthe materials constitutive model is no longer used. Through this new framework,materials data is directly combined with the general theorems of mechanics. Theobjective of this project is to develop a platform for data-driven computationalmechanics and apply it to analyze complex metal plasticity. In the first year, adata-driven computing mechanics platform and an elastic solver will beimplemented. Based on digital image correlation technology, materials dataidentification methods will be developed to obtain required materials data. Thesecond and third years will be focused on modeling and simulation of complexmetal plasticity. In the second year, solvers and materials data identificationmethods will be extended to inelasticity. In the third year, the connection ofmultiscale materials modeling will be explored to generate materials data bycrystal plasticity to analyze complex metal plasticity. The development of data-driven computational mechanics is still in its infancy. We hope that through thisthree-year effort, we can make leading contributions in this new field數據驅動計算力學材料大數據金屬塑性力學數位影像關聯Data DrivenComputational MechanicsMaterials Big DataMetalPlasticity,Digital Image Correlation