06/29/2021
By Sokny Long
The Francis College of Engineering, Department of Mechanical Engineering, invites you to attend a Master’s thesis defense by Jay Pandya on "Machine learning guided development of optimal padding concepts for combat helmets with improved blunt impact performance.”
MS Candidate: Jay Pandya
Date: Tuesday, July 6, 2021
Time: 1 to 2:30 p.m. (US EST)
Location: This will be a virtual defense via Zoom. Those interested in attending should contact Jay_Pandya@student.uml.edu and Advisor Murat_Inalpolat@uml.edu at least 24 hours prior to the defense to request access to the meeting.
Committee Chair: Murat Inalpolat, Ph.D., Associate Professor, Mechanical Engineering, UMass Lowell
Dissertation Committee Members:
- Scott Stapleton, Ph.D., Assistant Professor, Mechanical Engineering, UMass Lowell
- Jennifer Gorczyca, Ph.D., Associate Professor, Mechanical Engineering, SNHU
- Patrick Drane, Program Manager, Mechanical Engineering, UMass Lowell
Abstract:
The occurrence of Traumatic brain injuries (TBI) in the US has reached widespread proportions with well over 2 million new cases reported each year. The TBI can occur from a variety of threats including ballistic, blast, or blunt impacts. While ballistic and blast impacts have been extensively studied, blunt impacts and their mitigation have received considerably less attention for combat helmets. Therefore, this study is the first of its kind to focus on improving the design of the padded liner systems for the U.S. Army’s Advanced Combat Helmet (ACH) that reduces the risk of TBI for soldiers exposed to blunt impacts by exploring optimal design attributes and help reduce peak linear acceleration (PLA) experienced.
The main objective of this thesis is to develop optimal helmet padding concepts optimized for blunt impact performance. This study helps identify the best possible combination of pad material, geometry, position and weight along with strap tension for improved blunt impact performance of combat helmets subject to impacts up to 17 ft/s, applied at any location on the helmet shell.
To achieve this goal, a Gradient Boosting Decision Tree (GBDT) based surrogate model, which is a type of supervised machine learning algorithm, has been employed. This model trains on a series of limited finite element simulations of helmets undergoing impacts with the help of a Latin Hypercube sampling plan. This ensemble method has helped producing better results for helmets at low foam density and with less padding area as compared to the baseline version of the helmet. The standard criterion for a helmet with passing qualities against blunt impact is achieved if and when the acceleration of the headform does not exceed 150 g for all impact locations. Pad angles have been proved relatively more significant than foam factors and strap tension. Nape section, being the most critical location, performs significantly better at all recommended impact velocities while linear acceleration magnitudes are minimized for all of the other helmet sections.
The approach of using the surrogate model in conjunction with the gradient descent optimization algorithm has helped quantify the influence of design parameters and improved helmet liner design for blunt impact absorption. This design can be further optimized by considering the weight and comfort-related parameters that will move towards a more robust model. The proposed approach will help turn the current ad-hoc helmet design process into a data-driven design process, and explore design attributes for any military and/or civilian helmets for improved impact performance.
All interested students and faculty members are invited to attend the online defense via remote access.