Studi Penerapan Teknologi Wearable Device Untuk Mengawasi (Monitoring) Aktivitas Pengelasan FCAW (Flux-Cored Arc Welding)

Sunandar, Fransisco Juan (2022) Studi Penerapan Teknologi Wearable Device Untuk Mengawasi (Monitoring) Aktivitas Pengelasan FCAW (Flux-Cored Arc Welding). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 04111740000077-Undergraduate_tesis.pdf Restricted to Repository staff only Download (46MB) | Request a copy

Abstract

Teknologi pengelasan merupakan salah satu bagian yang tidak dapat dipisahkan dalam industri manufaktur bidang perkapalan. Welder adalah komponen utama dalam proses pengelasan tersebut, terutama aktivitas pengelasan itu sendiri. Salah satu penyebab terjadi keterlambatan pembangunan kapal adalah kinerja welder. Sementara fungsi pengawasan (monitoring) terhadap aktivitas pengelasan tersebut masih dilakukan secara manual, monitoring yang dilakukan berupa monitoring yang berbasis hasil las-lasan. Oleh karena itu penelitian ini dilakukan dengan tujuan supaya menerapkan teknologi wearable device untuk monitoring gerakan aktivitas dan kinerja welder pada pengelasan FCAW, alat kelengkapan utama yang digunakan untuk menyimpan data berupa smartphone, dan sebuah laptop untuk mengolah data aktivitas. Pertama, mengobservasi kondisi eksisting pengelasan FCAW. Kedua, dilakukan pengambilan data menggunakan alat wearable device MetaMotion, yang dilengkapi sensor accelerometer, gyroscope, dan magnetometer. Ketiga, melalui data hasil tersebut dianalisis dan dievaluasi menggunakan machine learning dengan algoritma Support Vector Machines (SVM) dan deep learning dengan algoritma Convolutional Neural Network-Long Short Term Memory (CNN-LSTM) untuk data aktivitas pengelasan FCAW dan kinerja juru las. Aktivitas welder yang dimonitoring terdiri dari pengelasan posisi 1G, 2G, 3G dan aktivitas lainnya. Hasil pelatihan data aktivitas pengelasan melalui SVM dengan beberapa variasi data memiliki nilai akurasi sebesar >75%, hasil pelatihan data kinerja welder melalui SVM dengan beberapa kriteria data memiliki nilai akurasi sebesar >85%. Hasil pelatihan data aktivitas pengelasan melalui CNN-LSTM dengan beberapa variasi data memiliki nilai akurasi sebesar >94%, pelatihan data kinerja juru las melalui CNN-LSTM dengan beberapa kriteria data memiliki nilai akurasi sebesar >96%.
===================================================================================================================================
Welding technology is an inseparable part of the manufacturing industry in the shipping sector. Welder is the main component in the welding process, especially the welding activity itself. One of the causes of delays in shipbuilding is the welder's performance, which is influenced by work attitude, and discipliness in time, duties and responsibilities. While the monitoring function of the welding activity is still carried out manually, the monitoring carried out is in the form of monitoring based on the results of the welds. Therefore, this research was conducted with the aim of applying wearable device technology to monitor the activity movement and welder performance in FCAW welding, the main equipment used to store data in the form of a smartphone, and a laptop to process activity data. First, observe the existing condition of FCAW welding. Second, data were collected using the MetaMotion wearable device, which was equipped with accelerometer, gyroscope, and magnetometer sensors. Third, through the data, the results are analyzed and evaluated using machine learning with the Support Vector Machines (SVM) algorithm and deep learning with the Convolutional Neural Network-Long Short Term Memory (CNN-LSTM) algorithm for FCAW welding activity data. The monitored welder activities consist of welding positions 1G, 2G, 3G and other activities. The results of training welding activity data through SVM with several variations of data have an accuracy value of >75%, the results of training welder performance data through SVM with several data criteria have an accuracy value of >85%. The results of training welding activity data through CNN-LSTM with several variations of data have an accuracy value of >94%, training welder performance data through CNN-LSTM with several data criteria has an accuracy value of >96%

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Sistem Monitoring, Welder, Pengelasan, Aktivitas, Kinerja, Wearable Device, Machine Learning, Deep Learning, Support Vector Machine, Convolutional Neural Network-Long Short Term Memory
Subjects:	V Naval Science > VM Naval architecture. Shipbuilding. Marine engineering
Divisions:	Faculty of Marine Technology (MARTECH) > Naval Architecture and Shipbuilding Engineering > 36201-(S1) Undergraduate Thesis
Depositing User:	Mr. Marsudiyana -
Date Deposited:	05 Nov 2025 01:37
Last Modified:	05 Nov 2025 01:37
URI:	http://repository.its.ac.id/id/eprint/128735

Actions (login required)

View Item