Built with AWS, LG CNS over six months

System unifies scattered data, enables natural-language search

A Korean Air Boeing 747-8i aircraft. [Provided by Korean Air]
A Korean Air Boeing 747-8i aircraft. [Provided by Korean Air]

Korean Air announced Friday that it has built a generative AI system capable of analyzing aircraft maintenance defects and retrieving similar past cases. The carrier plans to accelerate the digital transformation of its maintenance operations by consolidating millions of previously scattered maintenance records to support field technicians with faster information retrieval and better-informed decision-making.

Korean Air developed the new system over roughly six months in collaboration with LG CNS and Amazon Web Services. It uses generative AI to make maintenance data easier and faster to search and analyze.

Technicians can describe a defect symptom or situation in plain language, and the AI interprets the context and meaning to surface relevant maintenance histories and similar cases. The system also automatically converts abbreviations and unstructured field notes into standardized sentences, and extracts key information such as corrective actions taken and defect locations, improving the overall usability of maintenance data.

The system also consolidates what had been more than 90 separate database tables and millions of maintenance records into a single, unified interface.

Drawing on that consolidated data, the system provides statistics and trend analysis — including defect lifecycles by aircraft type and tail number, recurring defect patterns, and parts repair and action histories. Generative AI also analyzes correlations between maintenance data points and identifies defect occurrence patterns. Real-time flight messages received through air-to-ground communications can be queried alongside the maintenance data.

Korean Air expects the system to reduce the time engineers spend analyzing technical information and making decisions in the field, while improving the accuracy of engineering work. The system automatically preprocesses unstructured maintenance records entered by technicians, converting them into standardized data, and supports natural-language search so that past maintenance cases can be retrieved more quickly and easily.

"The new system was built to consolidate the vast maintenance data that had been scattered across different sources into one place, making it easier and faster to use on the maintenance floor with the help of AI," a Korean Air official said. "We will continue to accelerate our digital transformation based on advanced AI and data technology, enhancing our maintenance capabilities and further strengthening safe operations."


eyre@heraldcorp.com