Studi Penerapan Teknologi Wearable Device dengan Sensor Accelerometer, Gyroscope dan Magnetometer untuk Pengawasan (Monitoring) Pelatihan Pengelasan

Pramatama, Niko Kristian (2021) Studi Penerapan Teknologi Wearable Device dengan Sensor Accelerometer, Gyroscope dan Magnetometer untuk Pengawasan (Monitoring) Pelatihan Pengelasan. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

Text 04111740000008-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2023. Download (17MB) | Request a copy

Abstract

Pada industri manufaktur kapal di Indonesia masih dipergunakan proses pengelasan
manual dan semi otomatis (SMAW, GMAW,GTAW), sehingga juru las professional masih
sangat diperlukan. Proses pelatihan juru las professional dari tingkat dasar masih memerlukan
biaya yang cukup tinggi. Penelitian ini akan mencoba meningkatkan efisiensi dan efektifitas
proses pelatihan pengelasan dengan melakukan monitoring dan evaluasi secara real time dan
sistematis. Pertama dilakukan observasi kondisi eksisting pelatihan pengelasan dasar dan
pelatihan menuju profesional. Kedua, dilakukan record data dengan alat wearable device yang
dilengkapi sensor accelerometer, gyroscope dan magnetometer. Ketiga data hasil record
tersebut dianalisa dan dievaluasi menggunakan metode machine learning yaitu support vector
machine (SVM) untuk mengevaluasi tingkat kemampuan (skill level) peserta pelatihan.
Berdasarkan hasil analisa dapat diketahui kemampuan dan konsistensi peserta dalam
mempertahankan gerakan elektroda dengan membandingkan record data peserta dan instruktur.
Hasil dari sebaran data (scatter plot) dan juga matrix confusion dapat diketahui tingkatan skill
tiap peserta. Hasil training data didapat dengan nilai akurasi 88,2% untuk peserta pelatihan
dasar dan 77,8% untuk peserta pelatihan dari juru las biasa menuju ke juru las berkualifikasi
(qualified welder), dengan hasil uji seperti diatas, skenario proses pelatihan pengelasan
konvensional menggunakan elektroda dan spesimen sebagian dapat digantikan dengan latihan
gerakan tangan secara mandiri sampai menguasai teknik gerakan tangan dengan dimonitor dan
dievaluasi menggunakan wearable device. Setelah gerakan tangan memenuhi kriteria dapat
dilakukan evaluasi menggunakan spesimen dan electrode. Dengan skenario ini akan diperoleh
total penghematan biaya sebesar 35.30% atau sebesar Rp 31,000.00 per jam pelatihannya.
================================================================================================
In the ship manufacturing industry in Indonesia, manual and semi-automatic welding
processes (SMAW, GMAW, GTAW) are still used, so professional welders are still needed.
The process of training professional welders from the basic level still requires a fairly high cost.
This research will try to improve the efficiency and effectiveness of the welding training process
by monitoring and evaluating in real time and systematically. The first is to observe the existing
condition of basic welding training and professional training. Secondly, record data using
wearable devices equipped with accelerometer, gyroscope and magnetometer sensors. Third
data records were analyzed and evaluated using a machine learning method, namely support
vector machine (SVM) to evaluate the skill level of the trainees. Based on the results of the
analysis, it can be seen the ability and consistency of participants in maintaining electrode
movement by comparing the data records of participants and instructors. The results of the data
distribution (scatter plot) and also the confusion matrix can be seen by the skill level of each
participants. The results of the training data obtained with an accuracy value of 88.2% for basic
training participants and 77.8% for trainees from ordinary welders to qualified welders and with
the test results as above, the scenario of the conventional welding training process using
electrodes and specimens can be partially replaced with hand movement exercises
independently to mastering hand movement techniques by being monitored and evaluated using
wearable devices. After the hand movement meets the criteria, it can be evaluated using
specimens and electrodes. With this scenario, the total cost will be 35.30% or Rp. 31,000.00
per hour of training.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	support vector machine, machine learning, wearable device, welding training
Subjects:	V Naval Science > VM Naval architecture. Shipbuilding. Marine engineering > VM156 Naval architecture V Naval Science > VM Naval architecture. Shipbuilding. Marine engineering > VM298.5 Shipbuilding industri. Shipyards
Divisions:	Faculty of Marine Technology (MARTECH) > Naval Architecture and Shipbuilding Engineering > 36201-(S1) Undergraduate Thesis
Depositing User:	Niko Kristian Pramatama
Date Deposited:	03 Sep 2021 14:08
Last Modified:	03 Sep 2021 14:08
URI:	http://repository.its.ac.id/id/eprint/91611

Actions (login required)

View Item