Lestiawan, Junito Dwi Putra (2025) Sistem Sortir Pupuk Berdasarkan Kualitas Kode Produksi Berbasis Objek Deteksi Menggunakan Metode Convolutional Neural Network Pada Pengantongan Phonska V. Other thesis, Institut Teknologi Sepuluh Nopember.
![]() |
Text
2040211023-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (10MB) | Request a copy |
Abstract
Pada proses pengantongan pupuk Phonska di area Phonska V PT Petrokimia Gresik, sering ditemukan produk dengan kode produksi yang cacat atau bahkan tidak memiliki kode sama sekali. Penelitian ini bertujuan mengembangkan Sistem sortir pupuk berdasarkan kualitas kode produksi berbasis objek deteksi dengan menggunakan kamera dengan metode Convolutional Neural Network pada lini pengantongan Phonska V Departemen Pergudangan di PT Petrokimia Gresik. Perancangan Prototype sistem sortir ini bertujuan untuk mendemonstrasikan bahwa informasi dari deteksi kegagalan proses cetak kode produksi dapat digunakan sebagai Trigger untuk melakukan sortir pupuk yang tidak memenuhi standar. Selain itu penelitian ini juga bertujuan untuk melakukan pembuatan sistem deteksi realtime yang dimplementasikan pada Pengantongan Phonska. Sistem deteksi cetakan kode dibuat untuk mengetahui mendeteksi kegagalan dalam proses pencetakan kode produksi yang sering disebabkan oleh penyumbatan nozzle pada printer akibat debu. Produk yang memiliki kualitas cetakan kode produksi cacat ataupun tanpa kode akan terdeteksi oleh model YOLOv8 yang telah dibuat dan akan menyalakan Tower-lamp serta Buzzer untuk early Warning. Hasil Pengujian menggunakan 500 karung secara realtime menghasilkan nilai Precision 98,9%, Recall 92,6%, F1-Score 95,5%, Specificity 99,4%, Accuracy 92,6%, dan mAP 69,3%. Pengujian Prototype sistem sortir menggunakan sampel kode Good memiliki Accuracy 97,7%, dengan sampel Tidak-ada memiliki Accuracy 100% dan pada pengujian menggunakan sampel kode Bad, Accuracy mencapai 91,1%.
========================================================================================================================
During the Phonska bagging process in the Phonska V area of PT Petrokimia Gresik, products are often found with defective or missing production codes. This research aims to develop a fertilizer sorting system based on the quality of the production code. The system utilizes object detection with a camera and the Convolutional Neural Network (CNN) method on the Phonska V bagging line in the Warehousing Department at PT Petrokimia Gresik. The design of this Prototype sorting system is intended to demonstrate that information from detecting production code printing failures can be used as a Trigger to sort out fertilizer that does not meet standards. Additionally, this research aims to create a real-time detection system implemented at the Phonska bagging unit. The code print detection system was developed to identify failures in the production code printing process, which are often caused by nozzle blockages on the printer due to dust. Products with defective or missing production code prints will be detected by a custom-trained YOLOv8 model. This detection will activate a Tower-lamp and a Buzzer as an early Warning. The results from a real-time test using 500 bags yielded a Precision of 98.9%, a Recall of 92.6%, an F1-Score of 95.5%, a Specificity of 99.4%, an Accuracy of 92.6%, and a mAP of 69.3%. Prototype testing of the sorting system using Good code samples had an Accuracy of 97.7%, with samples without codes having an Accuracy of 100% and in testing using Bad code samples the Accuracy reached 91.1%.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Sistem Sortir, Convolutional Neural Network, Pengendalian Kualitas, Kode Produksi, Industri Pupuk, You Only Look Once |
Subjects: | Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) |
Divisions: | Faculty of Vocational > 36304-Automation Electronic Engineering |
Depositing User: | Junito Dwi Putra Lestiawan |
Date Deposited: | 04 Aug 2025 02:33 |
Last Modified: | 04 Aug 2025 02:33 |
URI: | http://repository.its.ac.id/id/eprint/124810 |
Actions (login required)
![]() |
View Item |