Framework for anomaly detection of flight-crew deviation from standard operating procedures: a data analytics approach.
| dc.contributor.advisor | King, Steve | |
| dc.contributor.advisor | Jennions, Ian K. | |
| dc.contributor.author | Igenewari, Vivian Rowoli | |
| dc.date.accessioned | 2024-07-31T11:30:03Z | |
| dc.date.available | 2024-07-31T11:30:03Z | |
| dc.date.issued | 2022-05 | |
| dc.description | Jennions, Ian K. - Associate Supervisor | |
| dc.description.abstract | Deviations from Standard Operating Procedure form a significant part of aviation incidents today involving loss of lives and other related costs. Previous work tailored towards detecting procedure deviations in flight operations have primarily been rule-based. The current method being used by airlines to detect operational, component fault and crew action anomalies within flight data is a rule-based Exceedance Detection technique which is only able to flag up known flight abnormalities. Lately, Anomaly Detection methods have been introduced to find, not just known, but unknown anomalies that deviate from the expected normal flight profile. There is a need to explore flight data using anomaly detection methods to detect subtle underlying misunderstandings of the flight crew in relation to deviations from laid down procedures which do not lead to incidents, under most conditions, or are hard to detect by the state-of-the-art method. However, these detection methods are limited in the type of anomalies they can find when implemented individually on heterogeneous flight dataset thereby missing critical anomalous flight incidents. In this work, Flight Data Recorder data of a fleet from a United Kingdom airline and a structurally similar publicly available dataset from the National Aeronautics & Space Administration are used. This study proposes a framework integrating an Ensemble anomaly detection technique (combining individual anomaly detection techniques into a single method) and a Case Based Reasoning system. The findings reveal that combining existing anomaly detection methods into an Ensemble can detect a wider variety of anomalies that were not flagged by individual methods. Also, the proposed reasoning design aims to filter for procedure deviations from the pool of anomalous incidents detected by the Ensemble. Detecting these procedure deviations is not just aimed at complementing crew training, improving procedures, and understanding automation design to put in place mitigation strategies but also to aid accident investigations by informing of accident flights with procedure deviations that may have been contributing factors. | |
| dc.description.coursename | PhD in Transport Systems | |
| dc.identifier.uri | https://dspace.lib.cranfield.ac.uk/handle/1826/22689 | |
| dc.language.iso | en | |
| dc.publisher | Cranfield University | |
| dc.publisher.department | SATM | |
| dc.rights | © Cranfield University, 2022. All rights reserved. No part of this publication may be reproduced without the written permission of the copyright holder. | |
| dc.rights | Attribution-NoDerivatives 4.0 International | en |
| dc.rights.embargodate | 2024-07-31 | |
| dc.rights.uri | http://creativecommons.org/licenses/by-nd/4.0/ | |
| dc.subject | Flight data | |
| dc.subject | Standard Operating Procedures | |
| dc.subject | Anomaly Detection | |
| dc.subject | Ensemble methods | |
| dc.subject | Machine Learning | |
| dc.subject | Data Analytics | |
| dc.subject | Case-Based Reasoning | |
| dc.title | Framework for anomaly detection of flight-crew deviation from standard operating procedures: a data analytics approach. | |
| dc.type | Thesis | |
| dc.type.qualificationlevel | Doctoral | |
| dc.type.qualificationname | PhD |