Anda di halaman 1dari 34

KLE DR.M.S.

SHESHGIRI
COLLEGE OF ENGINEERING AND
TECHNOLGY
U D YA M B A G , B E L A G AV I

ACCIDENT PREDICTION MODEL AND


MITIGATIVE MEASURES
Submitted by,

Imtiyaz Topinkatti 2KL15CV410


Sagar Khannukar 2KL15CV424
Asha kandre 2KL15CV404
Farheen Hangal 2KL15CV407

Under the guidance of,


Prof.H.G.Hunashikatti
CONTENTS

 INTRODUCTION
 OBJECTIVES
 METHODOLOGY
 DATA COLLECTION
 DATA ANALYSIS
 RESULTS AND OUTPUT
 CONCLUSION
 REFERENCES
INTRODUCTION
 Road Accidents cause high costs worldwide. Understanding why
accidents occur and how to prevent their re-occurrence is an
essential part of improving safety in any city.

 Belgaum city is unique, in terms of having two National Highways


passing within the city limits which causes huge of inter-city traffic.
The above factor further is coupled with the rapid urbanization,
with the intra-city vehicular traffic in the past few years having
increased considerably leading to severe stress on city's
transportation systems.

 The increase in vehicular traffic causes congestion and accidents


in the city due to heavy traffic load.
OBJECTIVES
 Identifying Independent Variables
 To understand the effects of different variables on accidents.

 Generating crash prediction model

 Generating accident severity model

 To suggest mitigative measures


METHODOLOGY
SELECTION OF SITE

DATA COLLECTION
(ACCIDENT
DATA,GEOMETRIC
FEATURES OF ROAD )

VOLUME COUNT ACCESS DENSITY MEDIAN DENSITY

VISSIM SPSS SOFTWARE


SOFTWARE

DATA INPUT DATA INPUT

GENERALISED LINEAR
SIMULATION IN VISSIM MODEL

OUTPUT / RESULTS OUTPUT/ RESULTS


DATA COLLECTION

 Volume count

 Accidental data of previous years & Black spot data

 Curvature & Rise and fall

 Median gap density

 Access points density

 Shoulder width
VOLUME COUNT FOR 1 HOUR
CBT to Ashoka Ashoka circle
Sl.no Type of Vehicle PCU value PCU value
circle toCBT
1 2 Wheeler 1320 660 944 472
2 3 Wheeler 356 284.8 344 275.2
3 Car 180 180 188 188
4 Bus 260 910 228 684
5 Truck 36 126 24 72
6 Tempo 12 12 12 12
7 LCV 124 272.8 60 132
8 Tractor Nil 0 8 8
9 Cycle 60 30 44 22

Ashoka circle to RTO to Ashoka


Sl.no Type of Vehicle PCU value PCU value
RTO circle
1 2 Wheeler 480 240 740 370
2 3 Wheeler 136 108.8 172 137.6
3 Car 80 80 228 228
4 Bus 92 322 140 420
5 Truck 36 126 80 240
6 Tempo 12 12 24 24
7 LCV 40 88 56 123.2
8 Tractor Nil 0 12 12
9 Cycle 16 8 12 6
ACCIDENTAL DATA OF PREVIOUS YEARS
& BLACK SPOT DATA
PHYSICAL CHARACTERISTIC OF ROADS
ACCESS
ROAD RISE/FALL CURVATURE MEDAIN SHOULDER LENGTH
POINT
SECTIONS (m/km) (degrees/km) GAPS WIDTH (m) (km)
DENSITY
CBT to
ASHOK 23.985 6.49 2 2.07 2 0.486
CIRCLE
ASHOK
CIRCLE to 31.105 25.34 2 2.96 7 0.737
RTO
RTO to
CHANNAMM 23.725 53.02 2 1.442 4 0.750
A CIRCLE
CHANNAMM
A CIRCLE to
20.544 8.17 2 1.83 7 0.925
BOGARVES
SIGNAL
BOGARVES
SIGNAL TO
18.117 8.45 4 2.36 14 1.540
RAILWAY
STATION
RAILWAY
STATION to 13.512 35.10 8 1.37 18 2.440
3RD GATE
3RD GATE to
19.523 32.19 2 2.825 6 1.280
BEMCO
BEMCO to
KLE ENGG. 22.026 22.04 ........ 2.625 4 0.630
COLLEGE
DATA ANALYSIS
VISSIM
 Vehicle input
Volume of vehicle taken for a particular stretch is
added to obtain the movement of vehicles in Vissim
model.
 Conflict points
Number of conflict points are identified then giving
preventive measures at particular locations.

 Vehicle Network Performance


In vehicle network performance by simulating in model
Behaviour of vehicle in network is recorded.
Ex: Average speed of the vehicles, Average Delay time
vehicle, stop average of vehicles etc.
SPSS SOFTWARE

GENERALISED LINEAR MODEL


 Generalised linear model can be use for the analysis of Accident
prediction model and as well as accident severity model.
 The equation generated in Generalised linear model by using
two variables
1) Dependent variable
 Number of accidents
2) Independent variable
 Curvature
 Median density
 Access Density
ACCIDENT PRIDICTION MODEL
 For Accident prediction model analyse of data is done by using
poissons log linear method or Negative binomial method.
 Accident prediction model equation is used for the prediction of
number of accidents that may occur in future.
CRASH SEVERITY MODEL
 For Crash severity model Binary logistic or Binary probit
method is used.
 This equation gives the accident occurring is Fatal or Non
Fatal.
Ex: 0 =Fatal & 1= Non Fatal
RESULT AND OUTPUT DATA
VISSIM OUTPUTS

CONFLICT POINTS
Reduction of conflict points by restricting the movement of vehicles at some
locations by closing median gaps or by making one way roads.

VEHICLE NETWORK PERFORMANCE


As the conflict points reduced there is change in vehicle network
performance results.
IMPROVEMENTS APPLIED
EXISTING REDUCED
ROAD CONFLICT CONFLICT LOCATION WHERE CONFLICT
STRETCH POINTS IN POINTS POINT REDUCED
NUMBERS NUMBERS
KLE TO BEMCO 24 NIL No action taken
BEMCO TO 3RD
7 NIL No action taken
GATE
Congress road to Ranapratap road
3RD GATE TO
78 68 & Opposite More at 1st gate (Both
STATION
by closing median gaps)
Restricting right turn to S
Telegraph road to Post office &
STATION TO
73 55 Restricting Right turn to Pension
BOGARVES
galli at camp area. (By closing
median gaps)
BOGARVES TO Closing traffic in front of Ganapati
9 7
CHENNAMA Temple
CHENNAMA TO Closing the right movement to the
16 12
RTO Civil judge court
RTO TO ASHOK Restricting the entry of traffic at
20 14
CIRCLE RTO office from CBT road
ASHOK CIRCLE Restricting the entry of traffic from
8 4
TO CBT CBT to Fort road
CHENNAMA TO RTO (EXISTING)
SAFETY
CONFLIC FRONT REAR MESO AVOID
VISIBILIT VISIBILIT DIST
T LINK1 LINK2 STATUS GAP GAP CRITICLE BLOCK
Y LINK1 Y LINK2 FACT
AREA:NO DEFAULT DEFAULT GAP MINOR
DEFAULT
1 100.0000 10002 100.00000 2 waits for 1
1 0.5 0.5 3.5 1.5 100.00%
1 100.0000 3 100.00000 1 waits for 2
2 0.5 0.5 3.5 1.5 100.00%
1 100.00000 10001 100.00000 1 waits for 2
3 0.5 0.5 3.5 1.5 100.00%
10002 100.00000 10005 100.00000 1 waits for 2
4 0.5 0.5 3.5 1.5 100.00%
6 100.00000 10006 100.00000 2 waits for 1
5 0.5 0.5 3.5 1.5 100.00%
6 100.00000 10007 100.00000 2 waits for 1
6 0.5 0.5 3.5 1.5 100.00%
6 100.00000 10008 100.00000 2 waits for 1
7 0.5 0.5 3.5 1.5 100.00%
1 100.00000 10009 100.00000 2 waits for 1
8 0.5 0.5 3.5 1.5 100.00%
6 100.00000 10009 100.00000 2 waits for 1
9 0.5 0.5 3.5 1.5 100.00%
1 100.00000 10010 100.00000 2 waits for 1
10 0.5 0.5 3.5 1.5 100.00%
6 100.00000 10010 100.00000 2 waits for 1 100.00%
11 0.5 0.5 3.5 1.5
3 100.00000 10011 100.00000 2 waits for 1
12 0.5 0.5 3.5 1.5 100.00%
1 100.00000 10012 100.00000 2 waits for 1
13 0.5 0.5 3.5 1.5 100.00%
1 100.00000 10013 100.00000 2 waits for 1
14 0.5 0.5 3.5 1.5 100.00%
6 100.00000 10013 100.00000 2 waits for 1
15 0.5 0.5 3.5 1.5 100.00%
6 100.00000 10014 100.00000 2 waits for 1
16 0.5 0.5 3.5 1.5 100.00%
CHENNAMA TO RTO (CHANGED)
SAFETY
CONFLIC VISIBILI FRONT REAR MESO AVOID
VISIBILIT DIST
T LINK1 LINK2 TY STATUS GAP GAP CRITICLE BLOCK
Y LINK1 FACT
AREA:NO LINK2 DEFAULT DEFAULT GAP MINOR
DEFAULT
1 100.00000 10002 100.0000 2 waits for 1
1 0 0.5 0.5 3.5 1.5 100.00%

1 100.00000 3 100.0000 1 waits for 2


2 0 0.5 0.5 3.5 1.5 100.00%

1 100.00000 10001 100.0000 1 waits for 2


3 0 0.5 0.5 3.5 1.5 100.00%

10002 100.00000 10005 100.0000 1 waits for 2


4 0 0.5 0.5 3.5 1.5 100.00%

6 100.00000 10006 100.0000 2 waits for 1


5 0 0.5 0.5 3.5 1.5 100.00%

6 100.00000 10007 100.0000 2 waits for 1


6 0 0.5 0.5 3.5 1.5 100.00%

6 100.00000 10008 100.0000 2 waits for 1


7 0 0.5 0.5 3.5 1.5 100.00%

1 100.00000 10009 100.0000 2 waits for 1


8 0 0.5 0.5 3.5 1.5 100.00%

6 100.00000 10009 100.0000 2 waits for 1


9 0 0.5 0.5 3.5 1.5 100.00%

3 100.00000 10011 100.0000 2 waits for 1


10 0 0.5 0.5 3.5 1.5 100.00%

1 100.00000 10012 100.0000 2 waits for 1


11 0 0.5 0.5 3.5 1.5 100.00%
CHENNAMA TO RTO EXISTING
CHENNAMA TO RTO REDUCED
VEHICLE NETWORK PERFORMANCE
CHENNAMA TO RTO
DELAY
DELAY STOP SPEED DIST TRAV TM DELAY STOPS DELAY VEH VEH DEMAN
COU TIME STOP DELAY
AVG S AVG AVG TOT TO TOT TOT STOPTO ACT ARR D
NT INT AVG LATENT
(ALL) (ALL) (ALL) (ALL) T(ALL) (ALL) (ALL) T (ALL) (ALL) (ALL) LATENT
(ALL)

1 0-100 1.875 0.000 42.26 0.000 11.90 1013.7 41.25 0 0.000 14 8 0.000 0.000

2 100-200 2.1370 0.043 42.984 0.0207 22.3186 1869.20 98.302 2 0.952 23 23 0.200 0.000

3 200-300 1.8818 0.000 43.242 0.0000 22.3106 1857.40 92.21 0 0.000 21 28 0.000 0.000

4 300-400 1.8396 0.000 42.179 0.0000 19.7870 1688.80 79.10 0 0.000 15 28 0.000 0.000

5 400-500 1.4576 0.000 42.666 0.0000 15.1832 1281.10 56.84 0 0.000 17 22 1.600 0.000
6 500-600 1.4870 0.000 43.576 0.0000 18.8190 1554.70 62.45 0 0.000 17 25 0.000 0.000

CHENNAMA TO RTO (CHANGED)


DELAY
DELAY STOP SPEED DIST TRAV TM DELAY STOPS DELAY VEH VEH DEMAN
COU TIME STOP DELAY
AVG S AVG AVG TOT TO TOT TOT STOPTO ACT ARR D
NT INT AVG LATENT
(ALL) (ALL) (ALL) (ALL) T(ALL) (ALL) (ALL) T (ALL) (ALL) (ALL) LATENT
(ALL)

1 0-100 1.62 0 42.27 0 11.9 1013.7 41.29 0 0 14 8 0 0

2 100-200 2.03 0 42.89 0 21.54 1858.6 93.29 0 0 24 22 0.2 0

3 200-300 1.60 0.04 41.7 0.34 22.25 1846.3 89.98 2 .862 21 29 0 0

4 300-400 1.27 0 42.23 0 18.64 1658.02 77.65 0 0 15 28 0 0

5 400-500 1.19 0 42.41 0 14.79 1194.6 54.87 0 0 16 22 1.6 0


6 500-600 1.33 0 43.55 0 17.8 1471.6 51.88 0 0 16 23 0 0
SPSS OUTPUT
• Accident Prediction Model
Model Information
Dependent Variable
Accident
Probability Distribution
Negative binomial (1)
Link Function
Log
Offset Variable
Exposure

Case Processing Summary


N Percent
Included
8 100.0%
Excluded
0 0.0%
Total
8 100.0%
• The Continuous Variable Information table can provide a rudimentary check
of the data for any problems The best you can get out of this table is to gain an
understanding of whether there might be over dispersion in your analysis

• Continuous Variable Information


Std.
N Minimum Maximum Mean Deviation
Dependent Accident
8 0 21 6.63 9.410
Variable
Covariate Curvature
8 6 53 23.85 16.199
Mediangap
8 0 8 2.75 2.375
Accesspoin
8 2 18 7.50 5.182
ts
Offset Exposure
8 9.35 13.30 10.4063 1.31185

• Determining how well the model fits


• The Goodness of Fit table provides many measures that can be used to
assess how well the model fits. However, we will concentrate on the value in the
"Value/df" column for the "Pearson Chi-Square" row, which is 1.108 in this
example, as shown below:
• Goodness of Fita
Value df Value/df
Deviance 6.354 4 1.588
Scaled Deviance 6.354 4
Pearson Chi-Square 3.848 4 .962
Scaled Pearson
3.848 4
Chi-Square
Log Likelihoodb -14.818
Akaike's Information
37.635
Criterion (AIC)
Finite Sample
Corrected AIC 50.968
(AICC)
Bayesian
Information 37.953
Criterion (BIC)
Consistent AIC
41.953
(CAIC)
Dependent Variable: Accident

Model: (Intercept), Mediangap, Accesspoints, Curvature, offset = Exposurea


a. Information criteria are in smaller-is-better form.
b. The full log likelihood function is displayed and used in computing information criteria.

• A value of 1 indicates equidispersion whereas values greater than 1 indicate overdispersion


and values below 1 indicate underdispersion.
Parameter Estimates
95% Wald
95% Wald Confidence
Confidence Interval for
Std.
Parameter B Interval Hypothesis Test Exp(B) Exp(B)
Error
Wald
Lower Upper Chi- df Sig. Lower Upper
Square
(Intercept) - - 4.374E 2.078E
1.5545 -6.991 41.692 1 .000 .001
10.037 13.084 -5 -6
Mediangap
.233 .8571 -1.446 1.913 .074 1 .785 1.263 .235 6.775
Accesspoint
.532 .4529 -.355 1.420 1.381 1 .240 1.703 .701 4.136
s
Curvature
-.274 .0950 -.460 -.088 8.307 1 .004 .760 .631 .916
(Scale)
1a
(Negative
1a
binomial)

Dependent Variable: Accident


Model: (Intercept), Mediangap, Accesspoints, Curvature, offset = Exposure
a. Fixed at the displayed value.

• This table provides both the coefficient estimates (the "B" column) of the Negative
Binomial regression and the exponentiated values of the coefficients (the "Exp(B)"
column).
Final result obtained from Accident prediction model

Coefficients

Intercept -10.037

Mediangap 0.233

Accesspoints 0.532

Curvature -0.274

• From regression analysis output we obtain coefficients for parameters and obtained
equation is:

Y = e −10.037+ 0.233×MedianDensity + 0.532×Accessdensity − 0.274×Curvature


SPSS OUTPUT
• Crash Severity Model
Model Information
Dependent Variable
Accidenta
Probability Distribution
Binomial
Link Function
Logit
Offset Variable
Exposure

a. The procedure models .00 as the response, treating 1.00 as the reference category .
Case Processing Summary

N Percent
Included
29 100.0%
Excluded
0 0.0%
Total
29 100.0%
• Categorical Variable Information

N Percent
Dependent Accident .00
9 31.0%
Variable
1.00
20 69.0%
Total
29 100.0%

• Continuous Variable Information


Std.
N Minimum Maximum Mean Deviation
Covariate Accessden
29 7.00 18.00 12.9655 4.70169
sity
Medianden
29 2.00 8.00 4.6897 2.57881
sity
Curvature
29 8.17 35.10 17.5431 12.96308
Offset Exposure
29 10.07 11.06 10.4921 .43274
• Omnibus Testa

Likelihood Ratio Chi-Square df Sig.


.485 2 .785

Dependent Variable: Accident


Model: (Intercept), Mediangap, Accesspoints, Curvature, offset = Exposurea
a Compares the fitted model against the intercept-only model.
Parameter Estimates

95% Wald
95% Wald
Confidence
Confidence Hypothesis Test
Interval for
Interval
Std. Exp(B)
Parameter B Exp(B)
Error Wald
Chi-
Lower Upper df Sig. Lower Upper
Squar
e
- - 3.529 1.545
(Intercept) 1.5961 -7.123 41.254 1 .000 .001
10.252 13.380 E-5 E-6
Accessden
-.145 .2847 -.703 .413 .259 1 .611 .865 .495 1.512
sity
Medianden
.174 .5143 -.834 1.182 .115 1 .735 1.190 .434 3.262
sity
Curvature 0a . . . . . . 1 . .

(Scale) 1b
Dependent Variable: Accident
Model: (Intercept), Mediangap, Accesspoints, Curvature, offset = Exposure
a. Fixed at the displayed value.

• This table provides both the coefficient estimates (the "B" column) of the Negative Binomial regression
and the exponentiated values of the coefficients (the "Exp(B)" column).

Final result obtained from Accident prediction model


Coefficients

Intercept -10.252

Mediangap -.145

Accesspoints .174

Curvature 0

• From regression analysis output we obtain coefficients for parameters and obtained equation is:

e −10.252− 0.145×MedianDensity + 0.174×Accessdensity + 0×Curvature


Y=
1 + e −10.252− 0.145×MedianDensity + 0.174×Accessdensity + 0×Curvature
CONCLUSION
 Based on the result obtained from the case studies It can be
concluded that by reduction of conflict points the number of
accidents can be reduced as the volume of traffic also decreases.

 The equations generated by accident prediction model gives


the number of accident will occur in future.

 The equations generated by accident severity model gives the


Fatal or Nonfatal accident.
REFERENCES
1. Siddharth S M, Gitakrishnan Ramadura “Calibration of VISSIM for
Indian Heterogeneous Traffic Conditions” 2nd Conference of
Transportation Research Group of India 2013
2. Brhane gebretensay, jayesh juremalani, “Road traffic accident analysis
and prediction model” International Research Journal of Engineering
and Technology, Volume: 05 Issue: 01
3. Jan K. Wachter , Patrick L. Yorio, “A system of safety management
practices and worker engagement for reducing and preventing
accidents”, Indiana University of Pennsylvania, Safety Sciences
Department, Johnson Hall Room 137, 1010 Oakland Avenue, 15705-
1063, United state.
4. Gulden Kaya Uyanik “A Study on Multiple linear regression analysis”
4th International Conference on New Horizons in Education 2014
5. Haikal Aiman Hartika , Mohd Zakwan Ramli, 2017 “Study of Road
Accident Prediction Model at Accident Black spot Area”. IJSRSET
Volume 3, Issue 5

Anda mungkin juga menyukai