HiDeNN-AI, LLC and Northwestern University (NU) propose to develop a data-driven physics-based modeling tool for evaluating mechanical properties of an as-built composite structure in this Phase I SBIR project. This goal is accomplished by integrating experimental data with data-driven modeling and analysis based on the Hierarchical Deep-learning Neural Network Artificial Intelligence (HiDeNN-AI) technology. HiDeNN-AI provides a structured methodology of mechanistic data science (MDS) that develops accurate reduced-order models at multiple scales by integrating data science with mechanistic principles discovered using neural network. To address the key challenges in evaluating the structure-property linkage in thick laminate composites, five technical objectives are proposed in this program for the Phase I base and optional period: i) Develop a computer vision tool (HiDeNN-I) to characterize experimental composite microstructures with defects, and their digital reconstruction for further analysis. (Phase I Base) ii) Extend our mechanistic reduced-order multiresolution clustering analysis (MCA) as HiDeNN-MCA for efficient multiscale modeling of the composite structure. (Phase I Base) iii) Demonstrate the capability of the HiDeNN-AI composite modeling tool on predicting local mechanical properties with uncertainties. (Phase I Base) iv) Develop a plan for validation and verification (V&V) of the proposed HiDeNN-AI software tool with the experiments. (Phase I Base) v) Extend the Phase I base for the elevated temperatures characterization, modeling, and demonstration. (Phase I Option) The key deliverables of the Phase I project include an add-on package to the major FE codes for evaluating mechanical properties of structural composites and its demonstration for simple coupon configurations. Future Phase II efforts will be directed towards application of the HiDeNN-AI tool for quantifying and certifying critical structural composites parts/components.
