Class of 2019: Karanpreet Singh uses artificial intelligence to accelerate the optimization of aircraft structures
As artificial intelligence becomes an increasingly important part of our daily lives, employers are looking to tech-savvy graduates to leverage this technology to their advantage. With the potential to reduce human effort and to produce more accurate and faster results, artificial intelligence is already rapidly changing a number of business sectors, including aerospace engineering.
Karanpreet Singh, who will graduate from Virginia Tech in December with a Ph.D. in aerospace engineering, is focused on using artificial intelligence to improve the structural design and optimization of aircraft structures. For aircraft designers, utilizing artificial intelligence will be key in designing structurally sound, fuel-efficient aircraft.
“I want my work to make an impact, not just on the aerospace community, but in many different disciplines like health, robotics, business, and education,” said Singh. “I hope to develop products which could have a positive impact on the people in developing countries.”
The structural makeup of aircraft of the future could mimic organic cellular structure. However, the design of new panels with arbitrary structures is computationally expensive, and applying the standard optimization methods used in the aerospace industry, known as finite element analysis, can take anywhere from a few days to months to complete.
In his research, Singh uses deep learning techniques to accelerate the optimization of various structures. Using deep neural networks, he has been able to approximate the buckling response when evaluating bio-inspired curvilinearly stiffened aircraft panels. These deep neural networks have accelerated the pace of standard methods by a factor of 200, or that the total time saved was 200 times. As an example, if the process typically took 100 hours, the use of artificial intelligence can reduce the time spent to half an hour. The deep learning techniques are performing at a 95 percent accuracy rate.
Singh’s research indicates that active learning has potential in speeding up the optimization of complex structures, like aircraft panels, heavy-duty truck chassis, submarine structures, and pressure vessels and pipes. By adaptively learning about the structure and improving its own accuracy during the optimization process, the use of artificial intelligence can reduce the required number of finite element analysis evaluations by more than 50 percent.
From Punjab, India, to Blacksburg, Virginia
Singh hails from Mohali, Punjab, India, and earned his undergraduate degree in mechanical engineering from Punjab Engineering College in Chandigarh, India, in 2014. As an undergraduate, he had the opportunity to complete a research internship at the Indian Institute of Technology in Delhi, India, and earned a fellowship for semester exchange at the Karlsruhe Institute of Technology in Karlsruhe, Germany. Both research experiences motivated him to pursue graduate studies at Virginia Tech.
After completing his undergraduate degree, Singh contacted numerous faculty with similar research interests across the globe for possible internship opportunities. During this time, he learned of Professor Rakesh Kapania and his group’s research activities at Virginia Tech. Singh spent one year as a visiting scholar within the Kevin T. Crofton Department of Aerospace and Ocean Engineering under Kapania, and his interest in the field of aerospace engineering grew.
“I was very impressed with his work-ethic, sincerity, and curiosity,” said Kapania, the Mitchell Professor of Aerospace and Ocean Engineering. “Being a mechanical engineer, he was not well-versed in thin-walled structures used in aerospace and ocean structures. But, he learned that material himself. I knew then that he would make a very good Ph.D. student, so I encouraged him to apply to the program.”
Kapania continued, “By his sheer hard work and can-do attitude, he has proven me right. After mastering machine learning, all by himself, he has successfully employed it to solve the very complex optimization problem of bio-inspired, stiffened plates and shells.”
Throughout his time at Virginia Tech, Singh has participated as a student member of the American Institute of Aeronautics and Astronautics, and the engineering honor society Tau Beta Pi. He is a two-time recipient of the John L. Pratt Fellowship and the Martin-Marietta Aircraft Fellowship. He has also had the opportunity to serve as a teaching assistant and felt great pride in his students’ successes.
He completed numerous internships as a doctoral candidate, including working as a machine learning research intern at Dassault Systèmes, in Waltham, Massachusetts, in summer 2019, and as a machine learning research intern at Qeexo Co. in Pittsburgh, Pennsylvania, in spring 2019.
As commencement approaches, Singh is looking for research positions related to artificial intelligence across the United States. Looking ahead to the future, he said, “I believe my journey on this path has just started. And for sure, Virginia Tech has helped me gain important values and skills for taking a step forward in the direction of my goals.”
— Story and photo by Jama Green