Nugroho, Wahyu (2026) Optimalisasi Golden Time Kondisi Abnormal Dengan Sistem Pendeteksi Api Dan Asap Menggunakan Algoritma YOLO Di PT. Pertamina Patraniaga Integrated Terminal Surabaya. Masters thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 6032242118-Master_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
6032242118-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy |
Abstract
Lamanya penanganan kecelakaan kerja khususnya kebakaran baik itu di Perkantoran, area Kilang Minyak, area terminal penimbunan bahan bakar minyak (BBM) menjadi penyebab kebakaran besar sehingga sulit untuk dipadamkan, apalagi jika berhubungan dengan bahan bakar minyak yg sifatnya mudah terbakar. Hal ini disebabkan karena proses pengawasan secara manual memiliki keterbatasan yaitu ketergantungan pada pengamatan manusia. Oleh karena itu, penelitian ini bertujuan untuk mengembangkan sistem deteksi otomatis api berbasis algoritma You Only Look Once (YOLO) yang diintegrasikan ke dalam platform berbasis (AI) Artificial intellegence. Penelitian ini menggunakan pendekatan computer vision dengan memanfaatkan data primer berupa visual api dan asap di lingkungan industri. Data visual dikumpulkan melalui dokumentasi langsung dan beberapa web sites sumber umum, kemudian dianotasi menggunakan Roboflow dan diproses melalui tahap pelatihan model YOLO dengan metode train-validation-test split. Evaluasi performa dilakukan menggunakan metrik precision, recall dan Mean Average Precision (MAP) dan F1Score. Sistem deteksi yang telah dikembangkan kemudian diimplementasikan ke dalam perangkat PC rakitan untuk memberikan kemudahan akses dan deteksi real time terhadap adanya sumber api. Evaluasi model dilakukan menggunakan metrik Precision, Recall, F1-Score, dan Mean Average Precision (mAP50). Hasil penelitian menunjukkan bahwa model terbaik menggunakan optimizer SGD, batch size 16, dan 150 epoch, dengan nilai Precision 84,1%, Recall 78,2%, F1-Score 81,0%, mAP50 83,7%. Sistem yang diimplementasikan mampu mendeteksi api dan asap secara real-time melalui CCTV pada berbagai kondisi pengujian sehingga mendukung percepatan golden time dalam penanganan kondisi abnormal di lingkungan PT Pertamina Patra Niaga Integrated Terminal Surabaya.
======================================================================================================================================
Delays in responding to workplace accidents specifically fires in office buildings, oil refineries, or fuel storage terminals often allow fires to escalate, making them difficult to extinguish, particularly when highly flammable fuels are involved. This issue stems from the limitations of manual monitoring, which relies heavily on human observation. Consequently, this study aims to develop an automated fire detection system based on the You Only Look Once (YOLO) algorithm, integrated into an Artificial Intelligence (AI) platform. The study employs a computer vision approach utilizing primary data consisting of visual imagery of fire and smoke within an industrial setting. Visual data was gathered through direct documentation and public online sources, annotated using Roboflow, and processed through the YOLO model training phase using a train-validation-test split method. Performance evaluation was conducted using metrics such as Precision, Recall, Mean Average Precision (mAP), and F1-Score. The developed detection system was implemented on a custom-built PC to facilitate easy access and real-time detection of fire sources. Model evaluation utilized Precision, Recall, F1-Score, and Mean Average Precision (mAP50) metrics. The results indicate that the optimal model utilized the SGD optimizer, a batch size of 16, and 150 epochs, achieving a Precision of 84.1%, Recall of 78.2%, F1-Score of 81%, and mAP50 of 83.7%. The implemented system is capable of detecting fire and smoke in real-time via CCTV across various testing conditions, thereby supporting faster golden times for managing abnormal situations at the PT Pertamina Patra Niaga Integrated Terminal Surabaya.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Deteksi Api, You Only Look Once, Computer Vision, Kebakaran, Roboflow, Fire and smoke Detection |
| Subjects: | Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) Q Science > QA Mathematics > QA76.9 Computer algorithms. Virtual Reality. Computer simulation. T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques. Image analysis--Data processing. |
| Divisions: | Interdisciplinary School of Management and Technology (SIMT) > 61101-Master of Technology Management (MMT) |
| Depositing User: | Wahyu Nugroho |
| Date Deposited: | 27 Jul 2026 15:01 |
| Last Modified: | 27 Jul 2026 15:01 |
| URI: | http://repository.its.ac.id/id/eprint/138629 |
Actions (login required)
 |
View Item |