Session: CIE-09-03 AMS/SEIKM: Artificial Intelligence and Machine Learning in Design and Manufacturing

Paper Number: 143998

143998 - Heterogeneous Transfer Learning for Design and Manufacturing Process Modeling Through Input Mapping and Multi-Fidelity Gaussian Process 

Data-driven modeling and optimization have become a cornerstone in several engineering design and modeling endeavors. The data-driven modeling frameworks enable faster optimization cycles and push the boundaries on the performance envelope of the system under study. Even with extended access to computational or manufacturing capabilities, the time\economic cost of evaluating design points are high as the simulation\experiment get closer to ground truth representation. The domain of transfer learning seeks to elevate this constraint on data requirements by leveraging data from other relevant or legacy data sets. However with the strong differentiation in design and manufacturing process, situations arise where the inputs of the target and source domain are heterogeneous in nature. Towards solving this problem, a heterogeneous transfer learning modeling framework is identified and implemented based on input mapping and multifidelity Gaussian processes. The framework is applied on three cases studies - (1) cantilever beam (2) ellipsoidal void and friction stir welding. The three case studies each represent a different facet of the design engineering community at large (1) Design Differentiated (2) Complexity Differentiated (3) Manufacturing Modality. Key insights on the input mapping transformation be-
tween the source\target domain domain and the performance of the integrated model. The results demonstrate adding source domain data facilitates a more accurate model.

Presenting Author: Sandipp Krishnan Ravi GE Aerospace Research

Presenting Author Biography: Dr. Sandipp Krishnan Ravi is a Research Engineer in the Probabilistics Design & Material Informatics group at GE Research, Niskayuna, NY. His current research areas include probabilistic machine learning and optimization methods in engineering applications. At GE Research, he is leading and supporting the technical development of novel methodologies, focusing on probabilistic modeling and optimization of material systems, manufacturing processes and component design for aerospace applications. He is currently serving as one of the special issue guest editor for ASME Journal of Computing and Information Science in Engineering. Sandipp received his PhD from University of Michigan.

Authors:

Sandipp Krishnan Ravi GE Aerospace Research
Piyush Pandita GE Aerospace Research
Changjie Sun GE Aerospace Research
Liping Wang GE Aerospace Research