Electric Arc Detection System Using the Random Forest Method on Electric Train Trolley Wires

Trains are a widely used transportation method for Indonesian people, with the use of overhead electricity as one of the highlights. This study focuses on sparks from overhead electricity, which can cause, that often occurs, a problem that is the focus of this study, because sparks from overhead electricity can cause various safety risks for very sensitive train. Therefore, early detection is very necessary considering that safety is the main concern of transportation. The development of a machine learning‐based system as an alternative is proposed for this problem. The use of various sensors that are relatively inexpensive on a small scale, such as ESP32 as a media device for data processing and PZEM‐017 sensors, can collect electricity data in real‐time. The dataset consists of 8000 samples (2000 samples per class) covering four operational conditions: no contact, normal, constant arc, and arc initiation. From these concerns, accurate output is obtained by calculating the average and standard deviation to see the results of the system used. In experiments with various configurations, each algorithm, such as Random Forest and SVM, has performed differently. Random forests show significant test results that outperform other methods. The results show that the level of accuracy in random forest reaches 99.50%, in addition, experiments were also carried out in the dynamic laboratory conditions, the system produced an accuracy of 95.60% so that these results can be trusted where the method used is not a perfect solution but to overcome the problem by combining the type of spark into an early warning for the system. This is distinct from the 99.56% average cross‐validation score obtained during hyperparameter tuning. In the future, this system will be tested in the railway area with the same method so that it can be used as a reference for implementation on trains.

Authors

Institutions

Publication Details

Journal
Applied Research
Published
2026-11-18
DOI
https://doi.org/10.1002/appl.70250
Primary Topic
Electrical Contact Performance and Analysis
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Electric Arc Detection System Using the Random Forest Method on Electric Train Trolley Wires

Wichai Siwakosit, Heru Syah Putra, Ahmad Sugiana
1 links
Applied Research
Electrical Contact Performance and Analysis
article

Electric Arc Detection System Using the Random Forest Method on Electric Train Trolley Wires

Wichai Siwakosit, Heru Syah Putra, Ahmad Sugiana
article en

Abstract

Trains are a widely used transportation method for Indonesian people, with the use of overhead electricity as one of the highlights. This study focuses on sparks from overhead electricity, which can cause, that often occurs, a problem that is the focus of this study, because sparks from overhead electricity can cause various safety risks for very sensitive train. Therefore, early detection is very necessary considering that safety is the main concern of transportation. The development of a machine learning‐based system as an alternative is proposed for this problem. The use of various sensors that are relatively inexpensive on a small scale, such as ESP32 as a media device for data processing and PZEM‐017 sensors, can collect electricity data in real‐time. The dataset consists of 8000 samples (2000 samples per class) covering four operational conditions: no contact, normal, constant arc, and arc initiation. From these concerns, accurate output is obtained by calculating the average and standard deviation to see the results of the system used. In experiments with various configurations, each algorithm, such as Random Forest and SVM, has performed differently. Random forests show significant test results that outperform other methods. The results show that the level of accuracy in random forest reaches 99.50%, in addition, experiments were also carried out in the dynamic laboratory conditions, the system produced an accuracy of 95.60% so that these results can be trusted where the method used is not a perfect solution but to overcome the problem by combining the type of spark into an early warning for the system. This is distinct from the 99.56% average cross‐validation score obtained during hyperparameter tuning. In the future, this system will be tested in the railway area with the same method so that it can be used as a reference for implementation on trains.

Applied ResearchVol. 5(6)
Kasetsart University (TH), Telkom University (ID)
Universitas Telkom
Industry, innovation and infrastructure
Electrical Contact Performance and Analysis
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon