Making better use of flight data from aircraft
Air transport is a very safe mode of transportation. Nevertheless, with the growing volume of air traffic and higher safety requirements, the aviation industry is being challenged to improve its safety record even further. To meet this challenge, the EU-funded 'Safety (and maintenance) improvement through automated flight data analysis' (SVETLANA) project was set up. The objective behind this EU–Russia cooperation was to improve the capabilities of flight data-monitoring programmes for civil aviation. Initially, the SVETLANA project developed a new analysis process. The first step used data-mining techniques in an automated detection process. This analysis was followed by a feedback loop whereby experts assessed smart maintenance and flight safety improvements. Several data-mining algorithms were also developed during SVETLANA. These developments were based on the most recent research in the field of anomaly detection, and patent applications have been filed for two of them. Project partners realised the architecture of the analysis system and the system's constraints were addressed. Here, the main innovations were layered above the definition of common formats, which is essential for standardisation of a common analysis process between Europe and Russia. Such formats enabled the aggregation of several external data sources and the development of improved data filtering techniques, parameters validation as well as innovative data flaws handling techniques. Finally, aircraft independent flight data analysis streamlined the development of patented anomaly detection algorithms (RDE) and dedicated HMI supported by End-users. An overall assessment of the new system for flight data analysis developed during SVETLANA was also carried out. This appraisal used both simulated and real-flight data from European and Russian sources. The anomaly detection algorithms developed during the now completed project will also be applied in areas other than the aviation sector. To date, they have been successfully utilised in the medical environment to develop a patient monitoring and prediction system in real time.
