Desain dan Implementasi Sistem Deteksi Anomali Getaran pada Motor Induksi Mesin Bubut untuk Mendukung Condition-Based Maintenance Berbasis Isolation Forest

Pratama, Aziz Yudha (2026) Desain dan Implementasi Sistem Deteksi Anomali Getaran pada Motor Induksi Mesin Bubut untuk Mendukung Condition-Based Maintenance Berbasis Isolation Forest. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2040221058-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (9MB) | Request a copy

Abstract

Motor induksi sebagai penggerak utama spindle mesin bubut beroperasi secara kontinu sehingga rentan mengalami degradasi komponen mekanis maupun elektris. Degradasi tersebut menyebabkan penyimpangan karakteristik getaran dari kondisi operasi normal yang didefinisikan sebagai anomali, sehingga berpotensi mengakibatkan downtime dan menurunkan efisiensi produksi. Pada klien PT Parametrik Solusi Integrasi, pemantauan kondisi mesin masih dilakukan secara manual dan bersifat reaktif tanpa kemampuan analisis maupun prediksi kerusakan. Hasil survei terhadap 25 unit mesin bubut menunjukkan bahwa meskipun setiap mesin menjalani 6 kali preventive maintenance setiap tahun, masih ditemui 2–3 kali kejadian downtime tidak terjadwal akibat permasalahan mekanis maupun elektris. Penelitian ini bertujuan merancang dan mengimplementasikan sistem untuk mendeteksi anomali berupa penyimpangan karakteristik getaran dari kondisi operasi normal motor induksi mesin bubut berbasis Isolation Forest. Data getaran pada axis X, Y, dan Z diproses melalui tahap preprocessing berupa pembersihan data dan normalisasi menggunakan StandardScaler sebelum dilakukan pelatihan model. Hasil validasi menunjukkan bahwa model Isolation Forest memperoleh nilai Accuracy, Precision, Recall, F1-Score, dan ROC-AUC masing-masing sebesar 92,32%, 75,47%, 64,61%, 69,62%, dan 96,9% pada Mesin Topas, serta 92,68%, 74,84%, 70,49%, 72,91%, dan 98,3% pada Mesin Krakatau. Selain itu, hasil perbandingan menunjukkan bahwa Isolation Forest memberikan performa terbaik dibandingkan Local Outlier Factor (LOF) dan One-Class SVM, dengan nilai Accuracy sebesar 92,32% pada Mesin Topas dan 99,10% pada Mesin Krakatau. Secara keseluruhan, sistem yang dikembangkan mampu melakukan akuisisi data, preprocessing, deteksi anomali, dan evaluasi model secara terintegrasi, sehingga dapat mendukung penerapan Condition-Based Maintenance (CBM) melalui deteksi dini anomali getaran pada motor induksi mesin bubut.
======================================================================================================================================
The induction motor, which serves as the main drive for a lathe machine spindle, operates continuously and is therefore susceptible to the degradation of mechanical and electrical components. Such degradation causes deviations in vibration characteristics from normal operating conditions, which are defined as anomalies and may indicate potential failures, leading to unplanned downtime and reduced production efficiency. At the client workshop of PT Parametrik Solusi Integrasi, machine condition monitoring is still performed manually and reactively without analytical or predictive capabilities. A field survey of 25 lathe machines revealed that although each machine undergoes preventive maintenance six times annually, 2–3 unplanned downtime events caused by mechanical and electrical failures still occur each year. This study aims to design and implement an Isolation Forest-based vibration anomaly detection system to identify anomalies characterized by deviations in vibration characteristics from the normal operating conditions of induction motors in lathe machines. Vibration data collected from the X-, Y-, and Z-axes were preprocessed through data cleaning and normalization using StandardScaler before model training. The validation results show that the Isolation Forest model achieved Accuracy, Precision, Recall, F1-Score, and ROC-AUC values of 92.32%, 75.47%, 64.61%, 69.62%, and 96.9%, respectively, for the Topas machine, and 92.68%, 74.84%, 70.49%, 72.91%, and 98.3%, respectively, for the Krakatau machine. Furthermore, a comparative evaluation demonstrated that Isolation Forest outperformed Local Outlier Factor (LOF) and One-Class Support Vector Machine (One-Class SVM), achieving Accuracy values of 92.32% for the Topas machine and 99.10% for the Krakatau machine. Overall, the developed system successfully integrates data acquisition, preprocessing, anomaly detection, and model evaluation, thereby supporting the implementation of Condition-Based Maintenance (CBM) through the early detection of vibration anomalies in induction motors used in lathe machines.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Getaran, Mesin Bubut, Motor Induksi, Downtime, Isolation Forest, StandardScaler, Deteksi Anomali, Condition-Based Maintenance, Preventive Maintenance
Subjects:	T Technology > TA Engineering (General). Civil engineering (General) > TA1573 Detectors. Sensors T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK4055 Electric motor T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105.546 Computer algorithms T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7878 Electronic instruments
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Aziz Yudha Pratama
Date Deposited:	06 Aug 2026 05:20
Last Modified:	06 Aug 2026 05:20
URI:	http://repository.its.ac.id/id/eprint/143723

Actions (login required)

View Item