Pendekatan Artificial Intelligence Dalam Mendeteksi Illegal Transhipment Kapal Berbasis Data Automatic Identification System

Puspitasari, Nita Wahyuni Dwi (2026) Pendekatan Artificial Intelligence Dalam Mendeteksi Illegal Transhipment Kapal Berbasis Data Automatic Identification System. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6019241016-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (18MB) | Request a copy

Abstract

Penelitian ini berfokus pada deteksi illegal transhipment dengan menggunakan data Automatic Identification System (AIS) daerah Batam tahun 2025. Variabel yang digunakan mencakup kecepatan kapal, jarak antar kapal, durasi pertemuan, dan perubahan arah heading, dengan indikator spesifik berupa pertemuan antara dua kapal dalam jarak kurang dari 500 meter, kecepatan di bawah 2 knot, heading saat sejajar dan head-on serta durasi pertemuan lebih dari 2 jam. Analisis dilakukan dengan menggunakan tiga metode machine learning yaitu Long Short-Term Memory (LSTM), Support Vector Machine (SVM), dan Gated Recurrent Unit (GRU). Metode LSTM menunjukkan performa terbaik dengan akurasi 99,15% pada data latih dan 91,2% pada data validasi, menjadikannya pilihan utama untuk deteksi transhipment. Sementara itu, metode SVM memperoleh akurasi sekitar 87,3%, tetapi kurang optimal dalam menangkap pola pergerakan dinamis yang kompleks. Meskipun SVM memiliki spesifisitas tinggi, kelemahan pada tingkat recall dapat menyebabkan deteksi kapal ilegal terlewat. Di sisi lain, GRU menawarkan efisiensi yang lebih baik untuk dataset terbatas, dengan akurasi tertinggi pada kategori kapal ikan sebesar 90,3%. Perbandingan ketiga metode menunjukkan trade-off antara kompleksitas model dan karakteristik data, di mana LSTM unggul dalam akurasi, GRU dalam efisiensi, dan SVM dalam klasifikasi yang layak. Hasil deteksi telah diintegrasikan ke dalam prototipe aplikasi berbasis web, dengan visualisasi peta interaktif yang menampilkan posisi kapal mencurigakan serta heatmap untuk mengidentifikasi daerah rawan. Solusi ini diharapkan dapat mendukung operasional otoritas maritim dengan memberikan data prediksi yang mudah dipahami.
====================================================================================================================================
This study focuses on the detection of illegal transshipment using data from the Automatic Identification System (AIS) from Batam 2025. The variables employed include vessel speed, distance between ships, meeting duration, and heading change, with specific indicators defined as encounters between two vessels within 500 meters, speeds below 2 knots, parallel and head-on headings, and meeting durations exceeding 2 hours. The analysis utilizes three machine learning methods such as Long Short-Term Memory (LSTM), Support Vector Machine (SVM), and Gated Recurrent Unit (GRU). The LSTM method demonstrates the best performance, achieving an accuracy of 99.15% on training data and 91.2% on validation data, making it the preferred choice for transshipment detection. In contrast, the SVM method yields an accuracy of approximately 87.3%, but is less effective in capturing complex dynamic movement patterns. Despite its high specificity, the lower recall rate of SVM may result in missed detection of illegal vessels. On the other hand, GRU offers better efficiency for limited datasets, achieving the highest accuracy of 90.3% in the fishing vessel category. A comparison of the three methods reveals a trade-off between model complexity and data characteristics, where LSTM excels in accuracy, GRU in efficiency, and SVM in reasonable classification performance. The detection results have been integrated into a web-based application prototype, featuring an interactive map that displays the positions of suspicious vessels and a heatmap to identify high-risk areas. This solution is expected to support maritime authorities by providing easily interpretable predictive data.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	AIS, Gated Recurrent Unit, Illegal transhipment, Long Short Term Memory, : AIS, Gated Recurrent Unit, Illegal transhipment, Long Short Term Memory, Support Vector Machine
Subjects:	T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models. V Naval Science > VK > VK555 Navigation. V Naval Science > VK > VK570 Optimum ship routing.
Divisions:	Faculty of Marine Technology (MARTECH) > Marine Engineering > 36101-(S2) Master Theses
Depositing User:	Nita Wahyuni Dwi Puspitasari
Date Deposited:	29 Jan 2026 07:36
Last Modified:	29 Jan 2026 07:36
URI:	http://repository.its.ac.id/id/eprint/130979

Actions (login required)

View Item