Advanced Machine Learning for Non-Destructive Testing

The Army depends on radiographic inspection for non-destructive testing (NDT) of munitions and armaments to identify defects before these products reach the warfighter. Highly skilled and experienced radiographers must interpret the inspection results. This process is not only laborious and time consuming, but also subjective and inconsistent. The development of artificial intelligence (AI) opens the opportunity to develop an automated system to identify potential defects to aid the interpretation of the results and expedite the inspection process. During Phase I, Sky Park Labs demonstrated DeepNDT, a suite of machine learning algorithms that assist a radiographer in the interpretation of radiographic inspection results by automatically identifying, quantifying, and visualizing potential defects and flaws. DeepNDT uses unsupervised and supervised deep learning algorithms for automatically detecting anomalies and highlights any deviations as potential defects. Then, the system classifies the defect type and characterizes its geometry and size for a radiographer to assess and make a final decision. These algorithms were evaluated on government-furnished data and various industrial datasets, achieving real-time performance and accuracy, precision, and recall rates of over 95% for detection and classification tasks. Phase II will focus on the development of all aspects of the technology into a fully functional prototype software tool integrated into the inspection workflow. The resulting technology has the potential to reduce the cognitive load on a level II NDT technician and thereby increase the efficiency and accuracy of the assessment of parts for defects.