Intelligent Automation of PAUT for Quantitative Defect Characterization and Evaluation

This project focuses on developing an automated decision making system for Phased Array Ultrasonic Testing (PAUT). Our goal is to reduce the industry’s reliance on manual defect evaluation. PAUT is standard in aerospace, nuclear, and manufacturing sectors for checking safety critical components. However, current practices depend heavily on a technician’s experience to interpret complex scan images, which often leads to variable results and slow inspections.

Our system automates this process by converting raw signals into high resolution B scan, C scan, and S scan views, integrating them into defect maps. We use advanced signal processing to filter noise and enhance clarity, ensuring the images are accurate and reliable. Furthermore, we characterize defects quantitatively using metrics like probability of detection and signal to noise ratio. This provides transparent evidence of flaw size, type, and location, reducing human error and ensuring consistency.

By automating these decisions, we eliminate the need for manual adjustments and make inspections more reliable. This approach not only shortens inspection times and cuts costs but also supports safer operations. Ultimately, combining PAUT with automation creates a scalable, dependable method for defect evaluation across the industry.

Faculty Supervisor:

Xihui (Larry) Liang

Student:

Partner:

Michigan State University

Discipline:

Engineering

Sector:

Advanced Manufacturing; Manufacturing and Construction; Artificial Intelligence

University:

University of Manitoba

Program: