Original scientific paper
https://doi.org/10.31306/s.66.2.3
An unsupervised learning-based analysis of the take-off behavior of the A320 and B738 at Sultan Hasanuddin international airport
Rizki Kurniati
; Faculty of Computer Science, Universitas Sriwijaya, Indonesia
*
Rossi Passerella
; Faculty of Computer Science, Universitas Sriwijaya, Indonesia
Indra Gifari Afriansyah
; Faculty of Computer Science, Universitas Sriwijaya, Indonesia
Osvari Arsalan
; Faculty of Computer Science, Universitas Sriwijaya, Indonesia
Aditya Aditya
; Faculty of Computer Science, Universitas Sriwijaya, Indonesia
Muhammad Rizki Fathan
; Faculty of Computer Science, Universitas Sriwijaya, Indonesia
Rani Silvani Yousnaidi
; Faculty of Computer Science, Universitas Sriwijaya, Indonesia
Harumi Veny
; Universiti Teknologi MARA (UiTM)
* Corresponding author.
Abstract
The purpose of this research was to look at the behavior of two well-known commercial aircraft types in Indonesia (the A320 and the B738) during the take-off phase. This was done to provide new information in the field of aviation, particularly flight safety. Observations were made at Sultan Hasanuddin International Airport by observing aircraft ADS-B data, which defines the behavior of the flight pattern. This ADS-B data is the subject of data analysis, which will subsequently be taught to the machine (computer) so that it can recognize the pattern and construct clusters. The purpose of this study is to utilize unsupervised learning, specifically K-Means clustering, to categorize and identify patterns in unlabeled ADS-B data obtained from AERO-TRACK.
To prepare the raw data and create a dataset, data analysis techniques were employed. The machine learning model generates three distinct clusters: cluster 1 represents aircraft take-off on two-thirds of the runway, cluster 2 represents aircraft take-off on the entire runway, and cluster 3 represents aircraft take-off on one-third of the runway. The elbow method is utilized to analyze and interpret the three clusters produced by the model. An interesting observation is that the B738 aircraft dominate in all three clusters, while the A320 aircraft dominate in clusters 1 and 3.
Notably, in cluster 2, there is a significant number of commercial planes taking off, accounting for 145 out of 628 flights. Based on the observed data spanning 91 days (September 26 to December 26, 2022), there is a 23% probability of runway excursion (overshooting the runway) in this cluster. Additionally, the research reveals that A320 aircraft demonstrate a safe zone take-off rate of 87%, whereas the B738 aircraft demonstrate a rate of 70.5%. These findings, derived from the analysis of ADS-B data such as GPS-Altitude and Coordinate, are intended to serve as valuable knowledge for aviation authorities, aviation users, and other stakeholders in the aviation industry.
Keywords
airport, data ADS-B, cluster, runway, K-Means
Hrčak ID:
318933
URI
Publication date:
8.7.2024.
Visits: 465 *