Data Analytics and Machine Learning Toolkit to Accelerate Materials Design and Processing Development

Navy has identified refractory high entropy alloy (RHEA) and metal additive manufacturing as two potential areas of interest. This includes designing new RHEA and optimizing metal additive manufacturing with specific material property requirements. Developing materials and processes via applying traditional experimentation and process optimization techniques is painfully slow due to the large number of variables in these systems. Therefore, application of machine learning (ML) techniques is envisioned. The objective of the STTR is to develop a) algorithms for transformation raw material and process data to extract useful information and b) software and modeling tools for guiding the development by automatically detecting data patterns and predicting material properties. Phase I aims demonstrating the proof-of-concept algorithm and tools to meet this objective. The proposed work includes collecting data from multiple sources, merging and extracting features, classifying the transformed data with unsupervised learning, developing predictive correlation with neural network and preliminary algorithms for material discovery and process optimization. The proof-of-concept software will be demonstrated for both RHEAs and metal AM process. In Phase II, the individual codes will be integrated into a GUI based easy to use software which can be applied by non-experts with minimal training