WASHINGTON, DC–What if you could tell that a plane’s brakes were about to fail based on the condition of the craft, age of the runway and current weather conditions? What if just such a plane was about to land at your airport? What if you could do something about it?
The William J. Hughes Technical Center Airport Technology Research and Development division—an office under the Federal Aviation Administration—has been collecting data on this problem and plans to build a machine learning algorithm that can predict brake failures before they happen.
But before the office can build that algorithm, it needs more data.
The Airport Technology Research and Development office maintains a database of incidents that occurred at or around airport reported through five federal agencies: the Aviation Safety Reporting System, National Transportation Safety Board, FAA Runway Safety Office Runway Incursion Database, Pilot Deviation System, and Vehicle/Pedestrian Deviation System.
But that data won’t be sufficient for a workable predictive analysis, according to a market survey released Monday. In order to build a reliable algorithm, FAA will need “access to a large volume of “aircraft landing data and the associated external contributing factors.”
The survey seeks feedback on the types of data available on:
- Raw aircraft data: Aircraft type, gross weight, flap position, touchdown ground speed, landing time and time zone, ICAO Airport Designator, and runway of landing.
- Processed aircraft data: Braking friction calculated from aircraft raw data for each third of the landing.
- External weather data: Precipitation type, precipitation intensity, air temperature, air pressure, humidity, pressure altitude, wind speed, and wind direction.
FAA is looking for at least two years of landing data—going back to at least July 1, 2017—from a variety of global airline carriers, landing conditions and plane conditions, including normal and degraded brakes.
