Projecte llegit
Títol: Statistical and predictive analysis of comercial aviation accidents and incidents
Estudiants que han llegit aquest projecte:
CARALPS BALLUS, PABLO (data lectura: 24-07-2026)- Cerca aquest projecte a Bibliotècnica
CARALPS BALLUS, PABLO (data lectura: 24-07-2026)Director/a: PONS PRATS, JORDI
Departament: FIS
Títol: Statistical and predictive analysis of comercial aviation accidents and incidents
Data inici oferta: 23-01-2026 Data finalització oferta: 23-09-2026
Estudis d'assignació del projecte:
- GR ENG SIST AEROESP
| Tipus: Individual | |
| Lloc de realització: EETAC | |
| Paraules clau: | |
| Aviation safety; Accident and incident reports; Normalised database; Commercial aviation; Bayesian predictive model; Contributing factors; Flight phase. | |
| Descripció del contingut i pla d'activitats: | |
| Mecanismos preventivos para mejorar los estándares de diseño de aeronaves,
El objetivo es desarrollar un modelo preventivo basado en el análisis de distintos parámetros e indicadores que permitan identificar tendencias relevantes y contribuir a mejorar la aeronavegabilidad y la seguridad en el diseño de aeronaves, basándose en la siguiente hipótesis: "Es posible anticiparse a una catástrofe, accidente o incidente aéreo mediante el uso de herramientas tecnológicas avanzadas y análisis predictivo." Pla de treball: - Estudi dels criteris de disseny d'aeronaus i certificat d'aeronavegabilitat - Estudi de les dades disponibles en les bases de dades obertes - Estudi de els tècniques de tractament de dades - Generació de la base de dades - Generació dels models basats en dades |
|
| Overview (resum en anglès): | |
| The objective of this study is to transform heterogeneous aviation accident and incident investigation reports into a consistent source of safety information and to develop a predictive model for identifying risk patterns in commercial aviation. To achieve this, the study focused on building a normalised database from official investigation reports and to estimate the probability that a reported occurrence results in an accident depending on its contributing factor and the flight phase in which it takes place.
A database of 952 commercial aviation occurrences from the 2010-2020 period was constructed using reports published by national and international investigation authorities. Each occurrence was normalised according to common criteria, including occurrence type, flight phase, contributing factor, aircraft consequences, human consequences, and the level of causal evidence available. The database was first analysed statistically and was then used to develop a Bayesian model that estimates accident probability for each combination of contributing factor and flight phase, together with its associated uncertainty. The results show that frequency alone does not determine safety relevance. Incidents represent 88.4% of the database, while the 11.6% classified as accidents concentrate all fatal and serious injury outcomes. Human and Technical factors appear with similar frequency, but Weather events show a much stronger association with accidents. The Bayesian model identifies Weather during the En route phase as the combination with the highest estimated accident probability, reaching 48.1%, compared with 3.0% for Technical factors during the same phase. The results also show that estimates based on fewer records present greater uncertainty and must therefore be interpreted with caution. Overall, the study demonstrates that official investigation reports can be transformed into a structured and useful source of knowledge for aviation safety. The resulting database and model provide a transparent method for identifying patterns, comparing scenarios, and supporting preventive safety decisions without replacing expert judgement or existing Safety Management Systems. |
|