Deteksi Dan Pemantauan Kapal Untuk Maritime Situational Awareness Menggunakan Data AIS Berbasis Kecerdasan Buatan

Pramudhita, Dhya Aqilla (2026) Deteksi Dan Pemantauan Kapal Untuk Maritime Situational Awareness Menggunakan Data AIS Berbasis Kecerdasan Buatan. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5018221014-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (15MB)

Abstract

Illegal, Unreported, and Unregulated (IUU) fishing atau penangkapan ikan ilegal, tidak dilaporkan, dan tidak diatur merupakan permasalahan penting dalam pengawasan kapal perikanan. Data Automatic Identification System (AIS) dapat digunakan untuk membaca perilaku kapal karena memuat posisi, waktu, kecepatan, arah gerak, dan identitas kapal secara berurutan. Namun, data AIS mentah masih dapat mengandung nilai kosong, duplikasi, loncatan posisi, jeda sinyal, serta pola gerak tidak wajar. Penelitian ini mengembangkan sistem deteksi kapal terindikasi IUU fishing menggunakan data AIS berbasis Long Short-Term Memory (LSTM). Tahapan penelitian meliputi preprocessing, feature engineering, pembentukan sequence trajectory, pelatihan model, validasi internal, dan pengujian external test. Sistem dibangun dalam empat pipeline, yaitu gear classification, spoofing detection, go-dark detection, dan transshipment candidate detection. Hasil pengujian eksternal menunjukkan nilai macro-F1 gear classification sebesar 0,758, F1 spoofing sebesar 0,752, event-F1 go-dark sebesar 0,651, dan event-F1 transshipment sebesar 0,750. Hasil tersebut menunjukkan bahwa model mampu mengenali pola perilaku dan anomali kapal dari rangkaian data AIS. Meskipun demikian, performa model masih dipengaruhi oleh perbedaan distribusi data, ketidakseimbangan kelas, kualitas sinyal, dan kemiripan pola antaraktivitas. Keluaran sistem digunakan sebagai indikator risiko awal untuk mendukung pemantauan dan investigasi, serta membantu petugas menentukan kapal yang memerlukan pemeriksaan lebih lanjut, bukan sebagai bukti hukum pelanggaran IUU fishing.
========================================================================================================================
============
Illegal, Unreported, and Unregulated (IUU) fishing is an issue in monitoring fishing vessels. Automatic Identification System (AIS) data can be used to analyze vessel behavior because they contain information on position, time, speed, heading, and identity. However, raw AIS data may contain missing values, duplicates, position jumps, signal gaps, and abnormal movement patterns. This study develops a system for detecting vessels involved in IUU fishing using AIS data and a Long Short-Term Memory (LSTM) model. Stages include preprocessing, feature engineering, sequence trajectory construction, model training, internal validation, and external testing. The system consists of four pipelines: gear classification, spoofing detection, go-dark detection, and transshipment candidate detection. External testing produced a macro-F1 score of 0.758 for gear classification, an F1 score of 0.752 for spoofing detection, an event-F1 score of 0.651 for go-dark detection, and an event-F1 score of 0.750 for transshipment candidate detection. These results show that the model can identify vessel behavior patterns and anomalies from AIS sequences. Nevertheless, performance is affected by data distribution differences, class imbalance, signal quality, and similarities between activities. The output is intended as an early risk indicator to support monitoring and investigation for further official assessment, not as legal evidence of IUU fishing violations.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	AIS, IUU fishing, LSTM, sequence trajectory, anomali kapal
Subjects:	V Naval Science > VK > VK555 Navigation. V Naval Science > VM431 Fishing boats
Divisions:	Faculty of Marine Technology (MARTECH) > Naval Architecture and Shipbuilding Engineering > 36201-(S1) Undergraduate Thesis
Depositing User:	Dhya Aqilla Pramudhita
Date Deposited:	30 Jul 2026 02:35
Last Modified:	30 Jul 2026 02:35
URI:	http://repository.its.ac.id/id/eprint/140230

Actions (login required)

View Item