Most airlines and MROs would be very happy if they could predict component breakdowns better, and indeed many OEMs, MROs and data scientists are moving in that direction. But even as this front is advancing, some are looking beyond prediction to using software to help techs fix the problems.
This is not as far-fetched as it sounds. Modern IT can do a variety of things better or at least faster than many humans. For example, Natural Language Processing could enable computers to search through digitized maintenance manuals much faster than techs, especially less-experienced techs, for just the right repair steps. And computers can ‘read’ and exploit massive amounts of unstructured data, including hand-written notes, that can help explain what may have gone wrong or what may go wrong in the future. Further, Machine Learning could benefit from the experience of many techs working on the same or similar problems to recommend the most likely trouble-shooting option, when several are available.
IBM calls use of these and related techniques “cognitive computing,” and, using its Watson system, has begun using it to help line technicians at Korean Air.
The airline had a huge volume of maintenance defect reports originally created by flight crews, cabin crews and techs. But these potentially very valuable reports were not standardized and very difficult to extract value from without administrative work that the airline simply did not have time for.
Watson had the power to digest these reports, including the initial problem and corrective actions taken. In the first ten months of 2016, the system wolfed down about a million defect reports and made sense of them for trouble-shooting and repair purposes. This approach reduced time for analyzing defect history by 90% and also cut actual trouble-shooting time. More than 2,000 Korean techs around the world now have access to all the knowledge accumulated by their colleagues in real time.