Theory Guided Machine Learning Process Modeling of in-situ Automated Fiber Placement

Current manufacturing processes for fiber-reinforced polymer composite materials are slow and rely heavily on manual intervention for quality control. The in-situ manufacturing of thermoplastic composites using Automated Fiber Placement (AFP) eliminates secondary thermal processing, which leads to decreased manufacturing times and cost. However, to reach an industrial implementation, we must improve our understanding of the non-linear and complex relationships that exist between all process parameters. Hence, optimization of the manufacturing process can be achieved by determining the relationships between process conditions (e.g., temperature, pressure, lay-up speed) and the final part quality. To address this knowledge gap, the proposed research project will develop a data-driven model using Theory Guided Machine Learning. The data-driven model will predict the temperature history achieved during AFP manufacturing using experimental data. The developed model can be exploited for more efficient control of the AFP process resulting in improved part quality and increased throughput.

Faculty Supervisor:

Farjad Shadmehri

Student:

Partner:

University of Washington

Discipline:

Engineering

Sector:

Education

University:

Concordia University

Program: