Implementasi Model Deep Q-Network untuk Deteksi Getaran dan Klasifikasi Kerusakan Rel pada Train-Based Monitoring System
DOI:
https://doi.org/10.32486/jeecae.v10i2.1136Abstract
Peningkatan keandalan infrastruktur perkeretaapian menuntut sistem pemantauan yang mampu mengenali pola getaran secara adaptif untuk mendeteksi kerusakan rel secara real-time. Penelitian ini menerapkan algoritma Deep Q-Network (DQN) sebagai agen cerdas tunggal dalam Train-Based Monitoring System (TBMS) untuk mengklasifikasikan jenis kerusakan rel berdasarkan pola getaran dari sensor bogie depan, bogie belakang, dan sensor inersia longitudinal. Dataset sintetis dibangun dari nilai Root Mean Square (RMS) amplitudo getaran yang mewakili energi vibrasi per segmen waktu. Hasil pengujian menunjukkan bahwa model DQN mencapai akurasi klasifikasi 87,8% dengan macro F1-score 0,875. Kinerja terbaik diperoleh pada kelas severe transverse crack (F1 = 1,000) dan defect (F1 = 0,999), sedangkan performa terendah terdapat pada kelas head crack (F1 = 0,522). Analisis Confusion Matrix ternormalisasi memperlihatkan dominasi diagonal utama dengan kesalahan klasifikasi rendah, menandakan kemampuan model dalam membedakan pola getaran secara presisi. Analisis SHapley Additive exPlanations (SHAP) menunjukkan bahwa sensor getaran bogie depan memberikan kontribusi tertinggi terhadap keputusan model sebesar 52,3%, diikuti oleh sensor IMU longitudinal 31,7%, dan sensor bogie belakang 16,0%. Hasil ini menegaskan bahwa pola getaran awal pada roda depan merupakan indikator paling signifikan dalam mendeteksi kerusakan rel, sedangkan dua sensor lainnya berperan sebagai sinyal pendukung untuk meningkatkan stabilitas prediksi model.
References
1] Jing L, Wang K, Zhai W. Impact vibration behavior of railway vehicles: a state-of-the-art overview. Acta Mechanica Sinica. 2021 Sep;37(8):1193–1221. doi:10.1007/s10409-021-01140-9
[2] Alvarenga, T. A., et al. (2021). Detection and classification system for rail surface defects based on eddy current. Sensors, 21(23), 7937. https://doi.org/10.3390/s21237937
[3] F. Flammini, C. Pragliola and G. Smarra, "Railway infrastructure monitoring by drones," 2016 International Conference on Electrical Systems for Aircraft, Railway, Ship Propulsion and Road Vehicles & International Transportation Electrification Conference (ESARS-ITEC), Toulouse, France, 2016, pp. 1-6, doi: 10.1109/ESARS-ITEC.2016.7841398.
[4] Ye, X. W., Su, Y. H., & Han, J. P. (2014). Structural health monitoring of civil infrastructure using optical fiber sensing technology: A comprehensive review. The Scientific World Journal, 2014, 652329. https://doi.org/10.1155/2014/652329
[5] Dang DZ, Lai CC, Ni YQ, Zhao Q, Su B, Zhou QF. Image classification-based defect detection of railway tracks using fiber Bragg grating ultrasonic sensors. Appl Sci (Basel). 2023 Jan 28;13(1):384. doi:10.3390/app13010384.
[6] Indu MA, Smitha ML. Enhancing the efficiency of damage detection in railway track using deep learning. Nanotechnology Perceptions. 2024;(20 S15):24–30. Available from: https://nano-ntp.com/index.php/nano/ article/view/4285
[7] Li, Y., Liu, J., & Wang, Y. (2016). Railway wheel flat detection based on improved empirical mode decomposition. Shock and Vibration, 2016, 4879283. https://doi.org/10.1155/2016/4879283
[8] Alisjahbana I. Using in-service train vibration for detecting railway maintenance needs. arXiv [Preprint]. 2024 May 3 [cited 2025 Mar 24]; Available from: https://arxiv.org/abs/2405.09560v1
[9] Hakim A.E, H. Hindersah and E. Rijanto, "Application of reinforcement learning on self-tuning PID controller for soccer robot multi-agent system," 2013 Joint International Conference on Rural Information & Communication Technology and Electric-Vehicle Technology (rICT & ICeV-T), Bandung, Indonesia, 2013, pp. 1-6, doi: 10.1109/rICT-ICeVT.2013.6741546.
[10] Hakim, A. E., Setyawan, S. B., & Junaedi, I. (2024). Intelligent Building Monitoring System for Earthquake-Resilient Structures using IoT and AI. Int. Res. J. Adv. Eng. Sci., 9(4), 62–67.
[11] PT Kereta Api Indonesia (Persero), Peraturan Dinas 10A: Perawatan Jalan Rel dengan Lebar 1.067 mm, Keputusan Direksi No. KEP.U/KL.104/XI/1/KA-2016, 11 Nov. 2016.
[12] Liu J, Chen S, Lederman G, Kramer DB, Noh HY, Bielak J, Garrett JH, Kovačević J, Bergés M. Dynamic responses, GPS positions and environmental conditions of two light rail vehicles in Pittsburgh. Scientific Data. 2019 Dec 1;6(1):146. doi:10.1038/s41597-019-0148-9.
[13] Bernal, E., Spiryagin, M., Santa, J., Toro, A., Wu, Q., Cole, C. (2025). Real-Time Digital Twin for Railway Systems. In: Huang, W., Ahmadian, M. (eds) Advances in Dynamics of Vehicles on Roads and Tracks III. IAVSD 2023. Lecture Notes in Mechanical Engineering. Springer, Cham. https://doi.org/10.1007/978-3-031-66971-2_47
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2025 JEECAE (Journal of Electrical, Electronics, Control, and Automotive Engineering)
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
