"Deformation Model Transfer via the Equivalent Effects of Lurking Variables in Additive Manufacturing"

ABSTRACT

Predictive models for geometric shape deformation constitute an important component in geometric fidelity control for additive manufacturing. However, the scope of application for any specific deformation model has traditionally been limited due to the wide variety of possible process conditions associated with different settings of lurking variables. We broaden the scope of deformation models by developing a novel framework for model transfer across different settings of lurking variables. Model transfer in our framework is formulated via the equivalent effects of lurking variables in terms of a base factor. The weakest sufficient condition on the data-generating mechanism in a new setting is identified that permits inference for the equivalent effects with respect to the mean. Bayesian methodology for modeling the equivalent effects and completing the model transfer are developed under this condition. Ultimately, our comprehensive approach connects different process conditions to provide a unified framework for geometric fidelity control in additive manufacturing.  

BIO

Arman Sabbaghi is an Assistant Professor in the Department of Statistics at Purdue University. His research interests include model building for improved quality control of complex engineering systems, Bayesian data analysis, experimental design, and causal inference. He received his PhD degree in Statistics from Harvard University in May 2014.

WEBSITE