Performance Enhancement of NDT Techniques Through AI
The integration of artificial intelligence (AI), machine learning, and advanced analytics with nondestructive testing (NDT) can significantly improve defect detection accuracy, accelerate inspection speed, and reduce reliance on the operator. Here, we highlight three recent patents reflecting current directions in AI-enabled NDT: intelligent ultrasonic weld evaluation, machine-learning-based acoustic imaging, and automated AI-driven inspection. Together, these patents illustrate how AI-based NDT is transitioning toward systems that automate data processing while improving detection reliability, inspection intelligence, and overall operational performance.
US2023/0228716A1
Comprehensive Real-Time Characterization of Ultrasonic Signatures from Nondestructive Evaluation of Resistance Spot Welding Process Using Artificial Intelligence
(Roman Gr. Maev, Donald Ryan Scott, Andriy Chertov, and Danilo Stocco)
The patent developers present an AI-driven system for real-time interpretation of ultrasonic signals collected during spot-weld inspection. The method uses trained machine learning models to characterize weld quality in situ by analyzing complex signal patterns. The system automatically classifies weld integrity and predicts quality metrics to identify potential defects such as weak bonding or structural inconsistencies. This patent represents a significant advancement, transforming conventional ultrasonic NDT into an intelligent in-line quality assurance tool that improves production speed, enhances consistency, and supports the transition toward smart automated manufacturing.
US12352727B2
Acoustic Imaging Techniques Using Machine Learning
(Benoit Lepage)
This patent focuses on the use of machine learning to improve acoustic imaging for inspection and defect detection. Acoustic images can contain substantial noise and distortion, complicating defect identification and reducing inspection reliability. The inventor intro[1]duces an AI-based model capable of enhancing image quality, suppressing unwanted noise, and extracting meaningful defect patterns that may not be easily recognized by human inspectors. The system can support faster interpretation by automatically highlighting regions of concern and improving visualization of cracks, voids, corrosion, delamination, and material discontinuities, thereby improving sensitivity, reliability, inspection confidence, and the overall efficiency of defect assessment processes.
US2023/0314373A9
System and Method for Automated Acquisition and Analysis of Electromagnetic Testing Data
(Philippe Mackay, Vincent Gaudreault, Florian Hardy, and Marco Michele Sisto)
The developers of this patent present an automated analytics framework for NDT systems that processes inspection data collected from probes using AI and machine learning algorithms. The system is designed to identify structural landmarks, detect potential defect indications, classify defect types, and generate inspection reports. By introducing data-driven intelligence into the inspection workflow, the patent reduces reliance on manual interpretation, improves consistency, and shortens analysis time, further supporting predictive maintenance strategies in industrial inspection environments.
Patents Roundup provides a review of recent patents of interest to the NDT community. If you’ve been granted a patent and would like to see it featured in an upcoming issue, please email Patents Editor Samir Mustapha at sm154@aub.edu.lb.
