Session: 06-02: Machine Learning and Statistical Methods in NDE II

Paper Number: 147625

147625 - Applying Machine Learning Methodologies for Identifying Varying Hidden Damage in Composite Plates Using Lamb Waves 

Abstract: 

Composites play a crucial role in the development of space structures due to their ability to maintain structural integrity while minimizing mass. The intricate composition of these materials is optimized for high performance, and the longevity of space vehicles relies heavily on the durability of composite assemblies. Detecting visible surface damage is relatively straightforward in assessing the safety of a vehicle for flight; however, assessing potential underlying damage requires Non-Destructive Evaluation (NDE) techniques, even when the exterior appears intact. Utilizing Ultrasonic Transducer (UT) inspection, differences between pristine and defective samples of thin composite plates become apparent through wave propagation. Machine learning (ML) techniques can then be employed to analyze various defect studies, allowing for the identification of potential patterns and quantification of damages. This study explores various ML methodologies, including regressions and classifications, to estimate the size, type (e.g., delamination versus disbond), and location of defects based on UT data. By quantifying waveform differences into dimensionless values, such as the Damage Index in NDE or Featured Extraction in ML, the significance of material properties and geometry diminishes. This ensures the development of a robust Convolutional Neural Network capable of being expanded and applied to thin plate inspections utilizing Lamb Waves, facilitating forecasting of defected regions with accuracy while only trained to a certain threshold.

Presenting Author: Gautham Viswaroopan University of California, Los Angeles

Presenting Author Biography: Gautham Viswaroopan is currently an Aerospace Engineering PhD student at UCLA in the Materials Degradation Characterization Laboratory focused on machine learning applications in non-destructive evaluation and structural health monitoring of composites. Gautham previously graduated from CU Boulder with his undergraduate studies in Mechanical Engineering and MS in Aerospace Engineering focused in structures and materials. Previous experiences include being a Matthew Isakowitz Fellow at The Aerospace Corporation; an intern at DigitalGlobe, NASA Ames, and Laboratory for Atmospheric and Space Physics (LASP); and a DoD Fellow through the X-Force Fellowship giving his overall work ranging from the commercial, private, and DoD sectors for spacecraft and launch vehicles. With his love for teaching as well, he has been a Teaching Assistant for graduate and undergraduate courses - presently in advanced solid mechanics. Outside of his career, Gautham is also a young-professional mentor for the Zed Factor Fellowship Program and actively involved in the largest student-led STEM organization, Students for the Exploration and Development of Space - a 501c3 non-profit to help provide projects, mentorship, and guidance to SEDS chapters globally as an international advisor. Gautham currently holds the title of an AIAA 20Twenties of 2020, recognizing top university talent for tomorrow's technology leaders in their 20's.

Authors:

Gautham Viswaroopan University of California, Los Angeles