Search for: All records

Active learning is a subfield of machine learning that focuses on improving the data collection efficiency in expensive-to-evaluate systems. Active learning-applied surrogate modeling facilitates cost-efficient analysis of demanding engineering systems, while the existence of heterogeneity in underlying systems may adversely affect the performance. In this article, we propose the partitioned active learning that quantifies informativeness of new design points by circumventing heterogeneity in systems. The proposed method partitions the design space based on heterogeneous features and searches for the next design point with two systematic steps. The global searching scheme accelerates exploration by identifying the most uncertain subregion, and the local searching utilizes circumscribed information induced by the local Gaussian process (GP). We also propose Cholesky update-driven numerical remedies for our active learning to address the computational complexity challenge. The proposed method consistently outperforms existing active learning methods in three real-world cases with better prediction and computation time.

 

Among metal additive manufacturing technologies, additive friction stir deposition stands out for its ability to create freeform and fully-dense structures without melting and solidification. Here, we employ a comparative approach to investigate the process-microstructure linkages in additive friction stir deposition, utilizing two materials with distinct thermomechanical behavior—an Al-Mg-Si alloy and Cu—both of which are challenging to print using beam-based additive processes. The deposited Al-Mg-Si is shown to exhibit a relatively homogeneous microstructure with extensive subgrain formation and a strong shear texture, whereas the deposited Cu is characterized by a wide distribution of grain sizes and a weaker shear texture. We show evidence that the microstructure in Al-Mg-Si primarily evolves by continuous dynamic recrystallization, including geometric dynamic recrystallization and progressive lattice rotation, while the heterogeneous microstructure of Cu results from discontinuous recrystallization during both deposition and cooling. In Al-Mg-Si, the continuous recrystallization progresses with an increase of the applied strain, which correlates with the ratio between the tool rotation rate and travel velocity. Conversely, the microstructure evolution in Cu is found to be less dependent on , instead varying more with changes to . This difference originates from the absence of Cu rotation in the deposition zone, which reduces the influence of tool rotation on strain development. We attribute the distinct process-microstructure linkages and the underlying mechanisms between Al-Mg-Si and Cu to their differences in intrinsic thermomechanical properties and interactions with the tool head.