Automakers could improve the quality of their welds by applying artificial intelligence to ultrasound analysis, according to a paper produced by a team from the Institute for Diagnostic Imaging Research.
The work earned lead author Vlad Tusinean kudos at the conference of the Canadian Institute for Non-Destructive Evaluation, held last week in Windsor.
Tusinean, a master’s student of computer science, was honoured for Best MSc Student Presentation. His project “Real-Time Outer Interface Characterization in Ultrasonic Images of Resistance Spot Welds Using Deep Learning” developed an approach that would allow manufacturers to modify the parameters during the welding process to prevent defects.
Doctoral student of mechanical, automotive, and materials engineering Maryam Shafiei Alavijeh finished second in the competition for best PhD student presentation for “Ultrasound-based inspection of butt-fused and electrofused medium-density polyethylene pipe joints.” Her research found that ultrasound evaluation supported by artificial intelligence can provide efficient, simple, inexpensive, and effective detection of defects in joints in plastic piping.
The recognition highlights the accomplishments of two very hardworking and talented graduate students, said physics professor Roman Maev, director-general of the Institute for Diagnostic Imaging Research and scientific program chair for the Non-Destructive Testing in Canada conference.
“The University of Windsor and the IDIR are proud of the contributions by these outstanding researchers, advancing our knowledge to the benefit of industry and our environment,” he said.
Dr. Maev also received an accolade at the conference, being elected a fellow of the Canadian Institute for Non-Destructive Evaluation.
Dean of science Chris Houser extended congratulations to Maev: “This is the latest recognition of Prof. Maev’s achievements and the impact that IDIR has had on industry and strengthening the reputation of the University of Windsor.”