Implementasi Edge Learning Classify Untuk Deteksi Kelengkapan Produk Sikat Dan Pasta Gigi Pada Lini Pengemasan Berkecepatan Tinggi

Babussalam, Babussalam (2026) Implementasi Edge Learning Classify Untuk Deteksi Kelengkapan Produk Sikat Dan Pasta Gigi Pada Lini Pengemasan Berkecepatan Tinggi. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2040221124-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Industri Fast-Moving Consumer Goods (FMCG) menuntut standar kualitas nol cacat pada lini pengemasan. Namun, inspeksi manual pada kecepatan 150 Product Per Minute (PPM) memiliki keterbatasan dalam mendeteksi ketidaklengkapan produk secara konsisten. Penggunaan sensor konvensional dengan titik deteksi tunggal juga terbukti tidak mampu mengakomodasi posisi kedua produk yang dinamis di dalam pocket. Proyek ini mengusulkan sistem inspeksi machine vision menggunakan algoritma Edge Learning Classify pada sebuah kamera cerdas yang diintegrasikan dengan Programmable Logic Controller (PLC) dan servo drive melalui jaringan EtherCAT. Sistem ini secara otomatis mendeteksi kelengkapan set sikat dan pasta gigi di dalam lima pocket konveyor sekaligus pada lini pengemasan berkecepatan tinggi 150 PPM. Proses pengambilan citra dieksekusi secara sekuensial berdasarkan sinkronisasi jarak pergerakan dari encoder motor servo. Kualitas citra visual sistem ini dimaksimalkan melalui penyesuaian parameter optik, yaitu penggunaan exposure time 1500 μs dan jarak lampu eksternal 35 cm, guna meminimalisir pantulan cahaya pada kemasan plastik. Untuk konfigurasi perangkat lunak, diterapkan penggunaan data latih seimbang dengan rasio 1:1 pada skenario Train Data 20-20 guna memastikan performa ekstraksi fitur yang terbaik. Hasil pengujian performa menggunakan metode confusion matrix menunjukkan bahwa sistem gabungan mencapai tingkat akurasi 97,60% dan F1-Score 98,21%. Selain itu, penerapan gerbang logika AND sebagai filter keamanan mutu sukses mencapai tingkat presisi 100%.
=====================================================================================================================================
The Fast-Moving Consumer Goods (FMCG) industry demands zero-defect quality standards on packaging lines. However, manual inspection at a speed of 150 Products Per Minute (PPM) has limitations in consistently detecting incomplete products. The use of conventional sensors with a single detection point has also proven incapable of accommodating the dynamic positions of both products within the pocket. This project proposes a machine vision inspection system utilizing the Edge Learning Classify algorithm on a smart camera, integrated with a Programmable Logic Controller (PLC) and a servo drive via an EtherCAT network. The system automatically detects the completeness of toothbrush and toothpaste sets across five conveyor pockets simultaneously on a 150 PPM high-speed packaging line. The image acquisition process is executed sequentially based on the movement distance synchronized from the servo motor encoder. The visual image quality is maximized through optical parameter adjustments, specifically utilizing an exposure time of 1500 μs and an external lighting distance of 35 cm, to minimize glare on the plastic packaging. For the software configuration, balanced training data with a 1:1 ratio is applied in the 20-20 Train Data scenario to ensure optimal feature extraction performance. Performance testing results using the confusion matrix method indicate that the combined system achieved an accuracy of 97.60% and an F1-Score of 98.21%. Furthermore, the application of an AND logic gate as a quality safety filter successfully achieved 100% precision.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Edge Learning Classify, Deteksi Kelengkapan, Lini Pengemasan Berkecepatan Tinggi, Edge Learning Classify, Completeness Detection, High-Speed Packaging Line.
Subjects:	T Technology > T Technology (General) T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK3070 Automatic control T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7871.674 Detectors. Sensors T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7882.P3 Pattern recognition systems
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Calrin Norika
Date Deposited:	10 Aug 2026 02:29
Last Modified:	10 Aug 2026 02:29
URI:	http://repository.its.ac.id/id/eprint/144257

Actions (login required)

View Item