Implementasi Convolutional Neural Network Untuk Deteksi Coil Dan Bara Api Pada Sistem Inspeksi Durasi Pembakaran Coil

Ongkosianbhadra, Gregorius Marcellinus (2025) Implementasi Convolutional Neural Network Untuk Deteksi Coil Dan Bara Api Pada Sistem Inspeksi Durasi Pembakaran Coil. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2040211049-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (54MB) | Request a copy

Abstract

Proses produksi obat nyamuk bakar memerlukan pengawasan ketat untuk memastikan durasi pembakaran coil sesuai standar masing-masing Stock Keeping Unit (SKU). Saat ini, proses inspeksi masih dilakukan secara manual, sehingga menghadirkan tantangan berupa keterbatasan pengawasan kontinu, volume data tinggi, serta risiko kesalahan input yang dapat memengaruhi kualitas produk dan Certificate of Analysis (CoA). Rata-rata terdapat 10 keterlambatan pengamatan per hari, dengan 197 data baru harian dan lebih dari 1.000 data bermasalah dalam setahun. Untuk mengatasi permasalahan tersebut, proyek ini mengembangkan sistem inspeksi semi otomatis berbasis Convolutional Neural Network (CNN) yang mampu menghitung dan mencatat durasi pembakaran coil secara akurat. Setelah melalui proses perancangan, implementasi, dan pengujian, sistem yang dibangun mampu mencatat durasi pembakaran secara semi otomatis dengan akurasi deteksi 98% untuk semua kelas dan akurasi perhitungan durasi pembakaran mencapat 95% pada konfigurasi optimal, yakni jarak kamera 25 cm dan pencahayaan 7,5 lux di laboratorium bakar PT X. Beberapa kendala teknis yang ditemukan meliputi false negative pada deteksi nyala bara api, false positive akibat perubahan bentuk coil setelah pembakaran panjang, serta gangguan visual seperti bayangan operator. Sistem ini telah terintegrasi dengan database SQL dan manajemen file video melalui REST API, mendukung digitalisasi dan dokumentasi hasil inspeksi. Dengan mengurangi ketergantungan terhadap inspeksi manual yang sebelumnya membutuhkan pengawasan hingga tujuh jam per coil, sistem ini berpotensi meningkatkan akurasi, efisiensi, dan efektivitas proses inspeksi. Proyek ini sekaligus menjadi studi kasus penerapan teknologi Computer Vision dalam industri manufaktur untuk menciptakan pengawasan kualitas yang modern dan berbasis data.
===============================================================================================================================
The production process of mosquito coils requires strict monitoring to ensure the burn duration of each coil meets the standard for its respective Stock Keeping Unit (SKU). Currently, the inspection process is still carried out manually, posing several challenges such as limited continuous supervision, high data volume, and the risk of input errors that may affect product quality and the Certificate of Analysis (CoA). On average, there are 10 delayed observations per day, with 197 new data entries daily and over 1,000 problematic records annually. To address these issues, this project developed a semi-automated inspection system based on Convolutional Neural Networks (CNN), capable of accurately measuring and recording coil burn durations. After undergoing the design, implementation, and testing phases, the developed system is capable of semi-automatically recording the burn duration with a detection accuracy of 98% across all classes and a burn duration calculation accuracy of 95% under the optimal configuration, a camera distance of 25 cm and lighting intensity of 7.5 lux in the burn laboratory of PT X. Several technical challenges were encountered, including false negatives in flame detection, false positives due to coil deformation after extended burning, and visual obstructions such as operator shadows. The system is integrated with an SQL database and video file management via a REST API, supporting the digitalization and documentation of inspection results. By reducing reliance on manual inspections that previously required up to seven hours of monitoring per coil, the system has the potential to significantly improve the accuracy, efficiency, and effectiveness of the inspection process. This project also serves as a case study for the application of Computer Vision technology in manufacturing, paving the way for modern, data-driven quality control.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Convolutional Neural Network (CNN), Durasi Pembakaran Coil, Sistem Inspeksi Semi Otomasi, Coil’s Burning Duration, Convolutional Neural Network (CNN), Semi-Automated Inspection System.
Subjects:	A General Works > AI Indexes (General) A General Works > AI Indexes (General) T Technology > T Technology (General) > T385 Visualization--Technique T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T58.5 Information technology. IT--Auditing T Technology > T Technology (General) > T58.6 Management information systems T Technology > T Technology (General) > T58.64 Information resources management
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Gregorius Marcellinus Ongkosianbhadra
Date Deposited:	07 Aug 2025 08:57
Last Modified:	07 Aug 2025 08:57
URI:	http://repository.its.ac.id/id/eprint/127965

Actions (login required)

View Item