Authors :
Hanum Arrosida; Agus Susanto; Adiratna Ciptaningrum; Tyan Rudianti; Masayu Nazar Surya Kencana; Rizal Mahmud
Volume/Issue :
Volume 8 - 2023, Issue 12 - December
Google Scholar :
http://tinyurl.com/495e2yku
Scribd :
http://tinyurl.com/mtss3b73
DOI :
https://doi.org/10.5281/zenodo.10432573
Abstract :
In Indonesia, trains are one of the most
popular means of transportation for Indonesians to help
with the mobility of passengers and goods. However,
train derailment is also something that happens quite
frequently. The train derailment was caused by several
factors, the rail line damage is one of the biggest
possible causes. For this reason, it is necessary to carry
out inspections on it to detect and find defects on the
rails for subsequent repairs. Manual inspections, as is
still often done by officers in Indonesia, have
shortcomings such as low efficiency, human error, and
danger. Automatic inspection can shorten inspection
time, reduce maintenance costs, and data can be real
time. The aim of this research is to create an automatic
inspection system using You Only Look Once (YOLO)
algorithm to rail line detect in Indonesia by taking case
studies in train operational areas along tracks that pass
through in two cities, namely from the Madiun Station
to West Station, in East Java Province. This area is
known as Daerah Operasi 7 (DAOP 7) with the 14 km in
distance. The result showed that the detection system
using the YOLO model had mAP value of 99.41%, a
precision value of 99%, a recall value of 99%, an f-score
value of 99%, and an average IoU value of 85.84%. The
YOLO model can detect railway track surface
abnormalities accurately and optimally. Therefore, it
can be used an automatic inspection for monitoring rail
line in Indonesia generally and rail line in East Java
Province, especially.
Keywords :
Train; popular means of transportation; rail line damage monitoring; DAOP 7 rail track; YOLO.
In Indonesia, trains are one of the most
popular means of transportation for Indonesians to help
with the mobility of passengers and goods. However,
train derailment is also something that happens quite
frequently. The train derailment was caused by several
factors, the rail line damage is one of the biggest
possible causes. For this reason, it is necessary to carry
out inspections on it to detect and find defects on the
rails for subsequent repairs. Manual inspections, as is
still often done by officers in Indonesia, have
shortcomings such as low efficiency, human error, and
danger. Automatic inspection can shorten inspection
time, reduce maintenance costs, and data can be real
time. The aim of this research is to create an automatic
inspection system using You Only Look Once (YOLO)
algorithm to rail line detect in Indonesia by taking case
studies in train operational areas along tracks that pass
through in two cities, namely from the Madiun Station
to West Station, in East Java Province. This area is
known as Daerah Operasi 7 (DAOP 7) with the 14 km in
distance. The result showed that the detection system
using the YOLO model had mAP value of 99.41%, a
precision value of 99%, a recall value of 99%, an f-score
value of 99%, and an average IoU value of 85.84%. The
YOLO model can detect railway track surface
abnormalities accurately and optimally. Therefore, it
can be used an automatic inspection for monitoring rail
line in Indonesia generally and rail line in East Java
Province, especially.
Keywords :
Train; popular means of transportation; rail line damage monitoring; DAOP 7 rail track; YOLO.
