AI Case Study

Korean Air reduces maintenance defect history analysis lead times by 90% and improves on-time performance with machine learning and natural language processing

Korean Air leverages IBM Watson to improve operational efficiency and on-time performance. Watson Explorer, Watson Natural Language Understanding and advanced content analytics are able to analyse vast amounts of data and aid maintenance crews at diagnosing and solving problems more efficiently. The airline manages to reduce its maintenance defect history analysis lead times by 90%.

Industry

Transportation

Airlines

Project Overview

"IBM Watson ingested structured and unstructured data from multiple sources including technical guidelines, non-routine logs, technician notes, inventory, trouble shooting time and material cost data, and in-flight incident history.

Watson Explorer, Watson Natural Language Understanding, advanced content analytics reveal previously hidden connections that help maintenance crews diagnose and solve problems more quickly, with more confidence.

Further, if an issue occurs in flight, the cabin crew can report it immediately to ground operations. Watson will access data from similar issues in the past and compare this information against technical guidelines including necessary materials and fixing time. Maintenance technicians fix the issue on the ground and enter their actions into the system to add to Watson’s knowledge.

With Watson, maintenance managers can also identify trends of issues in each season and can take these insights to the original equipment manufacturers for improvement.

The maintenance employee can now see patterns of defect and failure on equipment to make preventive maintenance allowing them to spend more time getting people places on time—and working to keep their 25 million passengers happy."

Reported Results

"Korean Air shortened its maintenance defect history analysis lead times by 90%. The maintenance employee can now see patterns of defect and failure on equipment to make preventive maintenance allowing them to spend more time getting people places on time—and working to keep their 25 million passengers happy."

Technology

Watson Explorer
Watson Natural Language Understanding
Advanced Content Analytics

"Built-in cognitive capabilities — powered by machine learning models, natural language processing and next-generation APIs — can unlock value from all your data and help you gain secure 360-degree view of customers, in context, to create better experiences for your customers."

Function

Operations

General Operations

Background

"Korean Air has years worth of historical maintenance records for hundreds of aircrafts in its fleet. But until recently, this vast amount of critical data was virtually unsearchable. That meant that maintenance technicians had to diagnose and fix issues without being able to tap into or interpret implications from valuable past learnings and courses of action."

Benefits

Data

Structured and unstructured data from multiple sources including technical guidelines, non-routine logs, technician notes, inventory, trouble shooting time and material cost data, and in-flight incident history.