AIS Field is thrilled to announce a significant milestone achieved last week, as the team successfully demonstrated and validated their first AI models in a real-world industrial environment. This marks a pivotal step in the development of their machine learning-based non-destructive testing (NDT) digital assistant. Over the past year, AIS Field has been diligently developing a digital assistant for NDT, utilising a decade’s worth of data from various industrial facilities across the globe. Through extensive research and development, the team crafted multiple neural network models employing deep learning techniques, benchmarking their performances to ensure top-tier results.
The recent validation study centred on eddy current tube inspections in two heat exchangers made from non-ferrous materials (ASTM B111 UNS C71500 and UNS S32100). During the data collection phase, the accuracy and speed of analysis and reporting were compared between a human operator and the AI model, known as “AIS X.” The results were groundbreaking: AIS X accelerated the analysis and reporting process by over 100%, with validation using a videoscope confirming a remarkable 99% accuracy in detecting and characterising defects. The AI model demonstrated its capability to analyse raw data and produce comprehensive reports within minutes. These successful validation results have invigorated the AIS Field team, which is now poised to integrate AI digital assistants into their robotic systems and conventional NDT operations. This integration aims to reduce human error while enhancing efficiency and speed significantly.
AIS Field is also preparing to document their approach in a scientific article titled “Development and Comparison of RNN, LSTM, and GRU Neural Network Models for Automated Eddy Current Inspection of Heat Exchanger Tubes Using Real World Industrial Data.” The article will delve into the methodologies and outcomes of their research. Interested parties can look forward to more details being shared at the upcoming BINDT 2024 and SPRINT Robotics WC2024 conferences. AIS Field’s advancements in AI-driven NDT signal a new era of efficiency and precision in industrial inspections, reinforcing their commitment to innovation and excellence in the field.
https://aisfield.com/
