Implementation Of A PSO-SVM Model For Single And Multi-Class Machine Fault Diagnosis Using Vibration Analysis

Manurung, Mirkline Samuel (2025) Implementation Of A PSO-SVM Model For Single And Multi-Class Machine Fault Diagnosis Using Vibration Analysis. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5007211017-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (6MB) | Request a copy

Abstract

Penelitian ini bertujuan untuk mengembangkan model sistem diagnosis yang efektif untuk mendeteksi dan mengklasifikasikan kerusakan mesin tunggal maupun multi-kelas, khususnya pada bearing dan shaft. Dengan memanfaatkan data getaran dari test rig eksperimental, model ini dirancang untuk mengurangi waktu henti produksi di era Industri 4.0. Penelitian ini menerapkan algoritma Support Vector Machine (SVM) yang hyperparameter-nya dioptimalkan menggunakan Particle Swarm Optimization (PSO). Metodologi yang diterapkan melibatkan ekstraksi fitur statistik dari sinyal getaran, diikuti oleh seleksi 9 fitur paling berpengaruh menggunakan Recursive Feature Elimination with Cross-Validation (RFECV). Hasil pengujian menunjukkan bahwa model yang dikembangkan mencapai performa sempurna dengan akurasi 100% pada 10 kelas kondisi mesin. Berdasarkan analisis perbandingan dengan standar ISO 10816-1, divalidasi bahwa performa tinggi ini dicapai pada kondisi kerusakan dengan tingkat keparahan yang signifikan (menengah hingga parah). Melalui implementasi strategi optimasi dua tahap, temuan dari penelitian ini diharapkan dapat memberikan kontribusi terhadap peningkatan efisiensi pemeliharaan mesin dan memberikan wawasan baru dalam penerapan teknologi diagnosis cerdas.
========================================================================================================================================
This study aims to develop an effective diagnostic system model to detect and classify single and multi-class machine faults, specifically in bearings and shafts. By utilizing vibration data from an experimental test rig, this model is designed to reduce production downtime in the Industry 4.0 era. This research implements a Support Vector Machine (SVM) algorithm whose hyperparameters are optimized using Particle Swarm Optimization (PSO). The applied methodology involves extracting statistical features from vibration signals, followed by the selection of the 9 most influential features using Recursive Feature Elimination with Cross-Validation (RFECV). Testing results show that the developed model achieved perfect performance with 100% accuracy across 10 machine condition classes. A comparative analysis against the ISO 10816-1 standard validated that this high performance was achieved on fault conditions of significant severity (medium to severe). Through the implementation of a two-stage optimization strategy, the findings of this research are expected to contribute to improving machine maintenance efficiency and provide new insights into the application of intelligent diagnostic technology.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Diagnosis Kerusakan Mesin, SVM, Vibrasi, PSO, Pembelajaran Mesin, Machine Fault Diagnosis, SVM, Vibration, PSO, Machine Learning
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA935 Vibration T Technology > TJ Mechanical engineering and machinery > TJ174 Maintenance and repair of machinery
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Mechanical Engineering > 21201-(S1) Undergraduate Thesis
Depositing User:	Mirkline Samuel Manurung
Date Deposited:	02 Aug 2025 15:06
Last Modified:	02 Aug 2025 15:06
URI:	http://repository.its.ac.id/id/eprint/125879

Actions (login required)

View Item