The use of machine learning in evaluating complex varied inspection data has made significant progress in recent years and continues to attract high research interest. Recent research mostly addresses automated defect detection using current procedures and techniques. With machine learning, new NDT methods with rich data and data fusion of multiple methods becomes viable, since the data analysis time and human capacity is no longer a limiting factor. This would allow fundamentally new data-driven inspections. The proposed research addresses this gap by developing rich multi-technique inspection data sets and developing novel data analysis techniques using machine learning to form new data-driven inspections. Current machine learning architectures are expanded and novel architectures explored to enable data fusion from varied sources.

2023

2027