Estimasi Remaining Useful Life untuk Sistem Traksi Pada Kereta Listrik

Puteri, Natasya Ariola Santoso (2025) Estimasi Remaining Useful Life untuk Sistem Traksi Pada Kereta Listrik. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6009231010-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Sistem traksi merupakan komponen vital dalam operasional kereta api karena berfungsi mengubah energi listrik menjadi tenaga mekanik. Untuk menjaga keandalan dan efisiensi operasional, pemeliharaan prediktif (Predictive Maintenance/PdM) menjadi strategi yang diutamakan karena mampu memperkirakan potensi kegagalan berdasarkan kondisi aktual komponen. Penelitian ini bertujuan untuk mengestimasi Remaining Useful Life (RUL) sistem traksi kereta dengan pendekatan data-driven menggunakan metode K-Nearest Neighbor (K-NN). Proses diawali dengan perhitungan Health Indicator (HI) dari parameter seperti arus dan daya. Nilai HI minimum ditentukan sebesar 0,5 sebagai batas kegagalan. Evaluasi dilakukan pada tiga skenario breakpoint (0,5; 0,7; dan 0,9), yang menunjukkan bahwa akurasi prediksi meningkat seiring bertambahnya data validasi, dengan selisih antara estimasi dan nilai aktual semakin kecil. Analisis Probability Density Function (PDF) menunjukkan peningkatan keyakinan model pada breakpoint tinggi. Hasil ini membuktikan bahwa pendekatan K-NN efektif dalam memodelkan degradasi dan memberikan estimasi RUL yang andal. Dengan demikian, strategi PdM berbasis data ini mampu meningkatkan efektivitas pemeliharaan dan mengurangi risiko kegagalan mendadak pada sistem traksi kereta api.
===================================================================================================================================
The traction system is a vital component in railway operations, responsible for converting electrical energy into mechanical power. To ensure operational reliability and efficiency, predictive maintenance (PdM) is prioritized as it enables the anticipation of potential failures based on the actual condition of components. This study aims to estimate the Remaining Useful Life (RUL) of a railway traction system using a data-driven approach with the K-Nearest Neighbor (K-NN) method. The process begins with the calculation of Health Indicators (HI) derived from key operational parameters such as current and power. A minimum HI value of 0.5 is defined as the failure threshold. Evaluation is conducted under three breakpoint scenarios (0.5, 0.7, and 0.9), revealing that prediction accuracy improves as the amount of validation data increases, indicated by a decreasing gap between estimated and actual RUL values. The analysis of Probability Density Function (PDF) also shows increased model confidence at higher breakpoints. These results demonstrate that the K-NN approach is effective in modeling degradation and providing reliable RUL estimations. Consequently, this data-driven PdM strategy contributes to enhanced maintenance effectiveness and reduced risk of unexpected failures in railway traction systems.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Health Indicator, K – Nearest Neighbor, Remaining Useful Life, Sistem Traksi,Traction System
Subjects:	T Technology > TL Motor vehicles. Aeronautics. Astronautics > TL152.5 Motor vehicles Driving T Technology > TL Motor vehicles. Aeronautics. Astronautics > TL220 Electric vehicles and their batteries, etc.
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30101-(S2) Master Thesis
Depositing User:	Natasya Ariola Santoso Puteri
Date Deposited:	07 Aug 2025 01:41
Last Modified:	07 Aug 2025 02:01
URI:	http://repository.its.ac.id/id/eprint/127865

Actions (login required)

View Item