Using Deep Learning to Automate the Detection of Flaws in Nuclear Fuel Channel UT Scans

TL;DR

This paper presents a CNN-based proof of concept for automating flaw detection in nuclear fuel channel ultrasonic scans, aiming to improve speed and accuracy over manual inspection methods.

Contribution

The study develops a CNN model trained on historical UT scan data to automatically identify flaw locations, reducing false positives and aiding manual flaw characterization.

Findings

CNN model successfully detects flaw locations in UT scans

Automated detection reduces inspection time compared to manual analysis

Model minimizes false positives in flaw identification

Abstract

Nuclear reactor inspections are critical to ensure the safety and reliability of a nuclear facility's operation. In Canada, Ultrasonic Testing (UT) is used to inspect the health of pressure tubes which are part of Canada's Deuterium Uranium (CANDU) reactor's fuel channels. Currently, analysis of UT scans is performed by manual visualization and measurement to locate, characterize, and disposition flaws. Therefore, there is motivation to develop an automated method that is fast and accurate. In this paper, a proof of concept (PoC) that automates the detection of flaws in nuclear fuel channel UT scans using a convolutional neural network (CNN) is presented. The CNN model was trained after constructing a dataset using historical UT scans and the corresponding inspection results. The requirement for this prototype was to identify the location of at least a portion of each flaw in UT scans while minimizing false positives (FPs). The proposed CNN model achieves this target by automatically identifying at least a portion of each flaw where further manual analysis is performed to identify the width, the length, and the type of the flaw.