Sistem Deteksi Kualitas Warna Pupuk Menggunakan Support Vector Machine (SVM) Pada Lini Produksi Gudang Urea

Putra, Nabil Andarsyah (2025) Sistem Deteksi Kualitas Warna Pupuk Menggunakan Support Vector Machine (SVM) Pada Lini Produksi Gudang Urea. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2040211045_Undergraduate_ Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Pengawasan parameter kualitas warna pupuk urea memainkan peran penting dalam memastikan standar mutu pada sektor produksi. Penelitian ini mengembangkan sistem deteksi kualitas warna pupuk berbasis algoritma Support Vector Machine (SVM) dengan fitur warna HSV (Hue, Saturation, Value). Sistem memanfaatkan kamera IP untuk menangkap citra pupuk secara real-time di jalur konveyor, yang kemudian diklasifikasikan ke dalam tiga kategori mutu: normal, pekat, dan pudar. Empat jenis kernel SVM dibandingkan, dan kernel Radial Basis Function (RBF) menunjukkan kinerja terbaik dengan akurasi 99,64%, macro F1-score 99,63%, precision-recall seimbang di atas 99% dan memperoleh mAP 99,82%. Sistem juga dilengkapi dashboard IoT untuk menampilkan hasil klasifikasi dan confidence score secara real-time. Hasil pengujian menunjukkan sistem mampu mendeteksi kualitas warna pupuk dengan akurasi tinggi dan konsistensi stabil dalam kondisi pencahayaan standar. Implementasi ini mampu mengurangi kesalahan inspeksi manual dan meningkatkan efisiensi proses produksi secara signifikan.
==========================================================================================================================================
Monitoring the color quality parameters of urea fertilizer plays a crucial role in ensuring quality standards in the production sector. This study developed a fertilizer color quality detection system based on the Support Vector Machine (SVM) algorithm with HSV (Hue, Saturation, Value) color features. The system utilizes an IP camera to capture real-time images of fertilizer on the conveyor line, which are then classified into three quality categories: normal, dark, and faded. Four types of SVM kernels were compared, and the Radial Basis Function (RBF) kernel demonstrated the best performance with an accuracy of 99.64%, a macro F1-score of 99.63%, a balanced precision-recall above 99%, and an mAP of 99.82%. The system also features a IoT dashboard to display the classification results and confidence scores in real-time. Test results show the system is capable of detecting fertilizer color quality with high accuracy and stable consistency under standard lighting conditions. This implementation can significantly reduce manual inspection errors and improve production process efficiency.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Support Vector Machine, HSV, deteksi warna pupuk, lini produksi, IoT, pengawasan kualitas,Support Vector Machine, HSV, fertilizer color detection, production line, IoT, quality control
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. T Technology > TS Manufactures > TS156 Quality Control. QFD. Taguchi methods (Quality control) T Technology > TS Manufactures > TS161 Materials management.
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Nabil Andarsyah Putra
Date Deposited:	21 Aug 2025 05:35
Last Modified:	21 Aug 2025 05:35
URI:	http://repository.its.ac.id/id/eprint/128153

Actions (login required)

View Item