Methodology development and implementation of a platform for non-destructive evaluation of both butt-fused and electrofused polyethylene pipe joints using ultrasound and deep learning

The infrastructure industry needs a way to non-destructively assess plastic pipe joints. We aim to research, design, and develop methodologies and a prototype device that allow automated inspection of these joints (made with the two most common joining methods) using ultrasound and artificial intelligence (AI). We propose a novel method that uses a two-element ultrasonic transducer (inexpensive but sensitive enough compared to previous methods). Ultrasonic signals produced by this transducer will be read by the AI, which will assess the joint (determine whether it contains a defect, and if so, characterize it in terms of type, size, and location), and display the results on a screen. Our proposed solution is inexpensive (so industry will adopt it), portable (as these joints are made usually in ditches as pipes are being inserted into the ground), and easy-to-use (hence, the AI component). The prototype will become a marketable product for our partner.

Intern: 
Majid Nezakat;Ryan Scott;Maryam Shafiei Alavijeh
Faculty Supervisor: 
Roman Maev
Province: 
Ontario
Sector: 
Partner University: 
Program: