International Journal of Scientific Research and Review ISSN NO: 2279-543X

Geographical Safety Approach to Railway Transportation

Using Machine Learning and Real Time Mapping.

Bhagavatula Roja Sri, Dr.Guru Kesava Das. Gopisettry

1Dept. Of CSE, 2HOD, Dept. Of CSE,
1,2Eluru collegeof Engineering,

Eluru, Andhra Pradesh, India.

Abstract: Railway transportation system is at high demand due to its reduced fair and longer
geographical coverage. World’s largest railway system lies in our country India. There are several
systems to avoid misfortunate incidences like accidents, rail blasts, and frequent problems on
railway network. The railway network is hugely dense so it is always difficult to find the prone
areas where the incidences take place frequently. Our proposed system is mainly a data collection
and processing system for misfortunate incidents of railway accidents. Anybody who saw the
accidents and can report to the railway. For that the system is created using JAVA which is
enabled to send mails from registered mail IDs to authorities. In the same website we can see the
overall graph of the accidents, accident numbers and accidents prone area where repeatedly
accidents occur, reason of accidents etc. Even latitude longitude based system is attached for exact
pin point tracing of accident details and location. It’s a JAVA based complete package to avoid
accidents and get complete report on a single place. Inspection time and tampering of facts are not
possible as all the data collection is done from genuine sources and eye witnesses.

Keywords: Railway transportation system, accidents, rail blasts, safety

1. Introduction

Safety has been one of the biggest concerns in the Indian Railways system. While the
number of accidents have gone down over the last few years, the number still remains
over 100 accidents a year. In light of the recent train accidents in Uttar Pradesh (UP), we
present some details around accidents and safety in the Indian Railways.

Causes of rail accidents

The number of rail accidents has declined from 325 in 2003-04 to 106 in 2015-16.The
number of rail accidents as per the cause are shown in the graph below. In 2015-16,
majority of the accidents were caused due to derailments (60%), followed by accidents at
level crossings (33%).1 In the last decade, accidents caused due to both these causes have
reduced by about half. According to news reports, the recent railway accidents in UP were
caused due to derailment of coaches

2. Existing system

General Information Collection Systems, In our existing systems the facts are totally
dependent on railway officials and inquiries. Whenever an accident happens several
officials come in action to find the cause of the accidents and number of victims. The
facts like why accidents happening on same spot, why same train getting accident, how
many lives exactly are over in the incident etc. are left over as this may cause several
departmental actions on officials. There is no individual and transparent system to know
the real truth.

Volume 7, Issue 1, 2018 6 http://dynamicpublisher.org/

International Journal of Scientific Research and Review ISSN NO: 2279-543X

3. Drawbacks of the existing systems

• Root cause and exact statistics are hidden.

• No independent body is involved in the inspection process so chance of evidence
tampering is likely to take place.
• No software based calculation system is designed yet and implemented for transparent
inspection.
• Geographical mapping of incidences on Google map is never used on a single place with
overall required data and real time mail based grievance system

4. Proposed System

Transparent Railway Safety System for Commoners, Our proposed system is a Java
based real-time approach to the solutions of all existing system drawbacks. It gave single
solutions to all the existing problems by providing transparent result. It is in the form of a
website in which any random person who was eyewitness of accident can make an
account and report accident. Even real-time grievance system is launched with toll-free
numbers where he can contact officials and send mail. The system is much efficient than
any other system implemented till date.

5. Advantages of the proposed systems

• All reporting and graphical statistics on a single place.

• Google map based exact location of accidents all over the country till date.
• Reason of accidents, number of accidents, same location accidents, number of casualties
all at a place.
• Genuine information based on eye witness and public.

Acknowledgement

Dr. Guru Kesava Das. Gopisettry is the Professor & HOD, in CSE Department, Eluru
college of Engineering Eluru. He obtained Doctorol Degree in Engineering Ph.D(CSE)
from Acharya Nagarjuna University, in the year 2014. He received his Master's Degree in
Engineering, M.E (CSE) from Anna University, Chennai in the year 2008. He completed
his Bachelor's Degree in Engineering, B.E (CSE) from Anna University, Chennai. He has
a rich experience of 17 years which includes Teaching, Research and Administration. He
served various reputed engineering colleges as a HOD for several years. He is a multi-
tasking personality and ago getter. His relentless efforts and commitment in Institutional
Administration lead few colleges to reach greater heights. He published several technical
papers in National and International Journals of repute. He is a certified professional of
IBM Rational.

Volume 7, Issue 1, 2018 7 http://dynamicpublisher.org/

International Journal of Scientific Research and Review ISSN NO: 2279-543X

References

[1] G. E. Hinton and R. R. Salakhutdinov, “Reducing the dimensionality of data with

neural networks.”
[2] A. Krizhevsky, I. Sutskever, and G. E. Hinton, “Imagenet classification with deep
convolutional neural networks
[3] C. Ekanadham, “Sparse deep belief net models for visual area v2,” Advances in
Neural Information Processing Systems
[4] W. Huang, G. Song, H. Hong, and K. Xie, “Deep architecture for traffic flow
prediction: Deep belief networks with multitask learning,”
[5] H. Larochelle, Y. Bengio, J. Louradour, and P. Lamblin, “Exploring strategies for
training deep neural networks
[6] Y. W. Teh and G. E. Hinton, “Rate-coded restricted boltzmann machines for face
recognition,”

Volume 7, Issue 1, 2018 8 http://dynamicpublisher.org/