Marc-Henri Bleu-Laine
(Advisor: Prof. Dimitri Mavris)

will propose a master’s thesis entitled,

A Methodology For The Prediction and Analysis of Precursors
to Flight Adverse Events

On

Wednesday, October 21 at 12:00 p.m.
Blue Jeans (https://bluejeans.com/589995189)

Abstract
Airplanes are known to be the safest means of transportation nowadays. The drastic improvements in aviation safety since its gain in popularity is undeniably a factor to the growth of the industry over the last several decades. This growth brought social and economic benefits throughout the whole world and was expected to keep its momentum pre-COVDI19. Stakeholders such as the National Aeronautics and Space Administration (NASA), the Federal Aviation Administration (FAA), the National Transportation Safety Board (NTSB), aircraft manufactures and airlines have developed systems, techniques, and technologies that are to thank for todays’ overall safety improvements and the reduction of accidents. The maintained growth of the industry is welcomed but current safety performances have been observed to stagnate instead of declining. With safety initiatives such as the (Flight Operational Quality Assurance) FOQA program and the growing number of data in aviation, many of the previous techniques used to understand the causes of accidents are not scalable. This led to the development of novel methods levering advanced analytical tools such as Machine Learning and Deep Learning.  However, current use cases have focused mainly on anomaly detection and system health monitoring, which does not bring enough reaction time to deal with an imminent event.

This research proposes the improvement of aviation safety through precursor mining. Precursors are defined as events that are highly correlated to the adverse event that they precede. Therefore, they provide predictive capabilities and can be used to explain predefined events. This thesis uses publicly available flight data to: 1) develop a novel Deep Learning method to identify and rank precursors of multiple adverse events, 2) use unsupervised learning algorithms to group flights based on their precursors to identify potential causes for these events at a fleet-level, and finally 3) use discovered precursors to the predefined events studied in this research to create a baseline from which precursors of potential events that are not predefined can be identified through the use of anomaly detection techniques.

Committee

• Prof.  Dimitri Mavris – School of Aerospace Engineering (advisor)
• Prof. Le Song – Computational Science and Engineering, College of Computing
• Dr. Tejas Puranik – School of Aerospace Engineering