Automated Geometric Shape Deviation Modeling for Additive Manufacturing Processes via Bayesian Neural Networks

A significant challenge in dimensional accuracy control of cyber-physical additive manufacturing systems (CPAMS) is the specification of geometric shape deviation models. The current practice of constructing tailor-made deviation models for each combination of computer- aided design model, additive manufacturing (AM) process, and process setting is impractical and inefficient for general application in CPAMS. We present a new framework and class of  Bayesian neural networks for automated and efficient deviation model building in CPAMS. The power and generality of our methodology is demonstrated with several case studies on both in- plane and out-of-plane deviation modeling for a wide variety of shapes manufactured under different stereolithography processes. Our framework yields models with comparable, and frequently better, predictive accuracy as existing, tailor-made deviation model building methods on the same sample size of manufactured shapes, but with dramatic improvements in computational efficiency, speed, and adaptivity. A distinct advantage of our method is that it can enable comprehensive and practical dimensional accuracy control for CPAMS without requiring detailed knowledge of the constituent AM processes. Ultimately, our new Bayesian neural network methodology facilitates geometric shape deviation modeling in general, and its broader impact is that it can help advance a smart cloud-based AM application with the potential of immediate practical application in CPAMS.