Sistem Klasifikasi Kondisi Motor Dust Collector Menggunakan Model Support Vector Machine (SVM) Guna Menunjang Condition-Based Maintenance

Winata, Nurdiwansyah Bagus (2024) Sistem Klasifikasi Kondisi Motor Dust Collector Menggunakan Model Support Vector Machine (SVM) Guna Menunjang Condition-Based Maintenance. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2040201128-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (6MB) | Request a copy

Abstract

Perusahaan Pestisida MSI merupakan salah satu perusahaan yang bergerak di bidang formulasi pestisida. Pada proses produksi, terdapat satu mesin penting yang menggunakan motor induksi 3 fasa sebagai penggerak utama, yaitu mesin dust collector. Namun, dalam dua tahun terakhir, terjadi sembilan kali kasus kerusakan pada mesin dust collector di MSI. Kerusakan tersebut terbagi menjadi beberapa jenis kerusakan, seperti kerusakan bearing, shaft misalignment, dan motor terbakar. Untuk mencegah kerusakan pada motor diperlukan penerapan metode maintenance yang efektif. Dengan data historis yang terbatas, metode Condition-Based Maintenance (CBM) adalah metode yang efektif. Dengan melakukan pemeliharaan hanya saat diperlukan berdasarkan kondisi aktual peralatan, CBM dapat mengurangi biaya yang tidak perlu dan mengurangi downtime yang tidak direncanakan. Dibantu dengan model Support Vector Machine (SVM), CBM pada proyek ini bertujuan untuk mengklasifikasikan kondisi motor secara aktual. Motor dengan kondisi normal, shaft miasalignment, dan bearing damage menjadi target klasifikasi pada proyek ini. Klasifikasi kondisi motor tersebut mengacu kepada data historis getaran dan suhu permukaan motor secara aktual. Dari penelitian yang telah dilakukan, pembacaan sensor HVT100 pada nilai getaran memiliki error sebesar 2,16% dan nilai suhu permukaan motor memiliki galat sebesar 3,89% dari 20 kali percobaan. Metode Support Vector Machine (SVM) menggunakan multiclass strategy OneVsOne (OVO) dan OneVsRest (OVR) dapat mengklasifikasikan kondisi motor normal, shaft misalignment, dan bearing damage secara aktual dengan akurasi dari 66% hingga 100%.
==================================================================================================================================
Pesticide company MSI is one of the companies active in the field of pesticide formulation. In the production process, there's one important engine that uses a three-phase induction motor as its primary driver, a dust collector. However, in the last two years, there have been nine cases of damage to the dust collection machine at MSI. Such damage is divided into several types of damage, such as bearing damage, shaft misalignment, and motorcycle burning. To prevent damage to the motorcycle, effective maintenance methods are required. With limited historical data, the Condition-Based Maintenance (CBM) method is an effective method. By performing maintenance only when necessary based on the actual condition of the equipment, CBM can reduce unnecessary costs and reduce unplanned downtime. Assisted by the Support Vector Machine (SVM) model, the CBM in this project aims to classify the motor condition in actual terms. Motorcycles in normal condition, shaft misalignment, and bearing damage are classified targets on this project. The classification of the condition of the motor refers to the historical data of the vibration and surface temperature of the actual motor. From the research that has been carried out the HVT100 sensor readings on the vibration value had an error of 2.16% and the surface temperature value of the motor had a error of 3.89% of 20 trials. Support Vector Machine (SVM) methods using multiclass strategies OneVsOne (OVO) and OneVsRest (OVR) can classify normal motor conditions, shaft misalignment, and bearing damage effectively with accuracy from 66% up to 100%.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Condition-Based Maintenance, Dust collector, Machine learning, Motor, Support Vector Machine
Subjects:	T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models. T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK2785 Electric motors, Induction.
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Nurdiwansyah Bagus Winata
Date Deposited:	03 Sep 2024 08:50
Last Modified:	03 Sep 2024 08:50
URI:	http://repository.its.ac.id/id/eprint/115583

Actions (login required)

View Item