Pengembangan Sistem Rekomendasi Tipe Pendekatan Pendaratan Pesawat Berdasarkan Kondisi Cuaca Dengan Pendekatan Supervised Learning

Putra, Tridiktya Hardani (2025) Pengembangan Sistem Rekomendasi Tipe Pendekatan Pendaratan Pesawat Berdasarkan Kondisi Cuaca Dengan Pendekatan Supervised Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5027211049-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Fase pendekatan dan pendaratan merupakan tahapan paling krusial dalam operasional penerbangan, di mana sebagian besar kecelakaan fatal terjadi. Petugas pengatur lalu lintas udara (ATC) dihadapkan pada tekanan tinggi dalam menentukan tipe pendekatan pendaratan yang tepat, terlebih dalam kondisi cuaca yang berubah-ubah. Penelitian ini mengembangkan sistem rekomendasi tipe pendekatan pendaratan berdasarkan data cuaca bandara secara real-time dengan menggunakan pendekatan supervised learning. Dataset didapatkan dari data D-ATIS (Digital Automatic Terminal Information Service) dan RVR (Runway Visual Range) enam bandara besar di Amerika Serikat melalui API. Algoritma yang digunakan adalah algoritma tree-based, yaitu Random Forest dan Extreme Gradient Boosting (XGBoost) serta kernelbased, yaitu Support Vector Machine (SVM) yang terdiri dari kernel linear, RBF, dan poly. Masing-masing diuji pada skenario fitur penuh dan sebagian, serta dikombinasikan dengan strategi meta-estimator OneVsRestClassifier. Model tree-based, baik pada skenario fitur penuh maupun fitur sebagian, hasil terbaik didapatkan oleh Random Forest. Pada skenario fitur penuh, model tree-based mendapatkan rata-rata macro precision, recall, dan F1-score terbaik sebesar 0,82; 0,92; dan 0,86. Sementara itu, pada skenario fitur sebagian, model tree-based memperoleh nilai rata-rata macro precision, recall, dan F1-score terbaik masing-masing sebesar 0,71; 0,79; dan 0,80. Hasil model kernelbased (SVC) terbaik pada fitur penuh didapatkan oleh kernel poly dengan rata-rata nilai precision, recall, dan f1-score terbaik masing-masing 0,75; 0,92; dan 0,79. Sementara itu, pada skenario fitur sebagian, model kernel-based memperoleh nilai rata-rata macro precision, recall, dan F1-score terbaik masing-masing sebesar 0,56; 0,73; dan 0,58; dengan menggunakan kernel RBF. Sistem juga berhasil diintegrasikan ke dalam aplikasi web interaktif untuk mendukung pengambilan keputusan oleh ATC.
============================================================================================================================================
The approach and landing phases are the most critical stages in flight operations, during which the majority of fatal accidents occur. Air traffic control (ATC) officers face intense pressure in determining the appropriate landing approach type, especially under changing weather conditions. This study develops a landing approach recommendation system based on real-time airport weather data using a supervised learning approach. The dataset is obtained from D-ATIS (Digital Automatic Terminal Information Service) and RVR (Runway Visual Range) data from six major airports in the United States via API. The algorithms used include tree-based methods such as Random Forest and Extreme Gradient Boosting (XGBoost), and kernel-based methods such as Support Vector Machine (SVM) with linear, RBF, and polynomial kernels. Each algorithm is tested under full and partial feature scenarios and combined with the OneVsRestClassifier meta-estimator strategy. In the tree-based models, Random Forest produced the best results under both the fullfeature and partial-feature scenarios. For the full-feature scenario, the tree-based models achieved the highest macro-average precision, recall, and F1-score of 0.82, 0.92, and 0.86 respectively. Under the partial-feature scenario, the tree-based models recorded the best macroaverage precision, recall, and F1-score of 0.71, 0.79, and 0.80 respectively. Among the kernelbased models (SVC) in the full-feature setting, the polynomial kernel delivered the top precision, recall, and F1-score averages of 0.75, 0.92, and 0.79 respectively. In the partialfeature scenario, the kernel-based models achieved their highest macro-average precision, recall, and F1-score of 0.56, 0.73, and 0.58 respectively using the RBF kernel. The system was also integrated into an interactive web application to support decision making by ATC.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Tipe pendekatan pendaratan, supervised learning, Random Forest, XGBoost, SVM, D-ATIS, Landing approach types, supervised learning, XGBoost, SVM, D-ATIS
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA336 Artificial Intelligence T Technology > T Technology (General) T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models. T Technology > T Technology (General) > T58.62 Decision support systems T Technology > TL Motor vehicles. Aeronautics. Astronautics > TL725.3.T7 Air traffic control
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Tridiktya Hardani Putra
Date Deposited:	10 Jul 2025 04:09
Last Modified:	10 Jul 2025 04:09
URI:	http://repository.its.ac.id/id/eprint/119482

Actions (login required)

View Item