Prediksi Penyebaran dan Pelacakan Kembali Kebakaran Hutan Menggunakan Model Fire Titans Transformer

Putra, Moch Fajar Aditya (2026) Prediksi Penyebaran dan Pelacakan Kembali Kebakaran Hutan Menggunakan Model Fire Titans Transformer. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002221084-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (18MB) | Request a copy

Abstract

Kebakaran hutan menjadi salah satu faktor berkurangnya pasokan oksigen secara signifikan. Seringnya penanggulangan kebakaran hutan kurang optimal yang utamanya disebabkan dari keterlambatan mendeteksi arah penyebaran kebakaran. Penelitian Tugas Akhir ini membangun sebuah model deep learning untuk mengestimasi penyebaran dan pelacakan kembali kebakaran hutan yang dinamakan Fire Titans Transformer (FiTT). Penelitian Tugas Akhir ini bertujuan untuk meramalkan penyebaran dan memprediksi pelacakan kembali kebakaran hutan dari input citra satelit hutan yang diberikan. Metode FiTT memiliki basis model S2R-FireTr dengan mengintegrasikan model Titans untuk mempelajari target area secara berkala. Metode FiTT diuji pada dataset publik Sim2Real. Hasil eksperimen menunjukkan bahwa FiTT berhasil meramalkan penyebaran dan memprediksi pelacakan kembali kebakaran hutan lebih akurat dibandingkan model baseline.
======================================================================================================================================
Wildfires are one of the factors significantly reducing the oxygen supply. Wildfire mitigation is often suboptimal, primarily due to delays in detecting the direction of fire spread. This research develops a deep learning model to estimate wildfire spread and backtracking, named Fire Titans Transformer (FiTT). This research aims to forecast the spread and predict the backtracking of wildfires from given forest satellite imagery input. The FiTT method is based on the S2R-FireTr model by integrating the Titans model to learn the target area over time. The FiTT method is tested on the public Sim2Real dataset. Experimental results show that FiTT successfully forecasts the spread and predicts the backtracking of wildfires more accurately than the baseline model.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Kebakaran hutan, Peramalan Penyebaran Kebakaran, Prediksi Pelacakan Kembali Kebakaran, Deep Learning. Wildfire, Wildfire Spread Forecasting, Wildfire Backtracking Prediction, Deep Learning.
Subjects:	Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Moch. Fajar Aditya Putra
Date Deposited:	04 Aug 2026 07:27
Last Modified:	04 Aug 2026 07:27
URI:	http://repository.its.ac.id/id/eprint/143598

Actions (login required)

View Item