Gaussian process tensor responses emulation for droplet solidification in freeze nano 3D printing of energy products

Freeze nano 3D printing is a novel process that seamlesslyintegrates freeze casting and inkjet printing processes. It canfabricate flexible energy products with both macroscale andmicroscale features. These multi-scale features enable goodmechanical and electrical properties with lightweight structures.However, the quality issues are among the biggest barriers thatfreeze nano printing, and other 3D printing processes, need tocome through. In particular, the droplet solidification behavioris crucial for the product quality. The physical based heattransfer models are computationally inefficient for the onlinesolidification time prediction during the printing process. In thispaper, we integrate machine learning (i.e., tensordecomposition) methods and physical models to emulate thetensor responses of droplet solidification time from the physicalbased models. The tensor responses are factorized with jointtensor decomposition, and represented with low dimensionalvectors. We then model these low dimensional vectors withGaussian process models. We demonstrate the proposedframework for emulating the physical models of freeze nano 3Dprinting, which can help the future real-time processoptimization.