26
Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent Transportation Systems, St. Louis, MO, USA, October 3-7, 2009 Ameena Padiath, Lelitha Vanajakshi, Shankar C. Subramanian, and Harishreddy Manda REPORTER: WEI HSU 2014/10/23 1

Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

Embed Size (px)

Citation preview

Page 1: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

Prediction of Traffic Density for Congestion Analysis under Indian

Traffic Conditions

Proceedings of the 12th International IEEE Conferenceon Intelligent Transportation Systems, St. Louis, MO,

USA, October 3-7, 2009Ameena Padiath, Lelitha Vanajakshi, Shankar C.

Subramanian, and Harishreddy Manda

R E P O RT E R : W E I HS U

2014/ 10/ 23

1

Page 2: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

2

Introduction The available infrastructure is not enough to meet the

demand. Solution: flyovers, road widened

External events have effect on traffic congestion. Solution: Intelligent Transportation System (ITS)

Page 3: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

3

Introduction (cont.) ITS control strategies take many forms

Metering flow onto roadways

Dynamically retiming traffic signals

Managing traffic incidents

Providing travelers with information

Page 4: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

4

Introduction (cont.) Traffic data were collected using videographic. (Manual)

Using loop detectors for collecting traffic data.

(Automatic)

Comparison of the performance Historic method

Artificial Neural Networks (ANN)

Model based approach

Page 5: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

5

Outline Data Collection Prediction Techniques Comparison of The Techniques Conclusion

Page 6: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

6

Outline Data Collection Prediction Techniques Comparison of The Techniques Conclusion

Page 7: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

7

Schematic Representation of The Study Site

Page 8: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

8

Data Collection Data extraction was carried out using video image

processing software.

Video data were collected

Initial number of vehicles inside the section

Number of vehicles entering and leaving the section

Page 9: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

9

Calculating Traffic Density

is traffic density

is average vehicle length

is detection zone length

is percentage occupancy time

Total occupancy time in time period T

Time period of observations (hours)

Page 10: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

10

Calculating Traffic Density (cont.)

Page 11: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

11

Calculating Traffic Density (cont.)Density(vehicles / lane-mile) %occ Flow Conditions

0 - 12 0 - 5 Free-flow operations

Uncongested flow conditions

12 - 20 5 - 8 Reasonable free-flow operations

20 - 30 8 - 12 Stable operations

30 - 42 12 - 17 Borders on unstable operations

42 - 67 17 - 28 Extremely unstable flow operations

Near-capacity flow conditions

67 - 100 28 - 42 Forced or breakdown operations Congested flow

conditions> 100 > 42 Incident situation

operations

Page 12: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

12

Outline Data Collection Prediction Techniques Comparison of The Techniques Conclusion

Page 13: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

13

Prediction Techniques Three different techniques can predict the traffic density

Historic technique

Artificial Neural Network (ANN) technique

Model based approach

Page 14: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

14

Historic Technique It is the most popularly adopted methods for short term

prediction

Historical average value will be used for prediction

Using average of the density data from four days’ data to

predict density for fifth day

Page 15: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

15

Artificial Neural Network(ANN) Technique ANNs can be trained to learn a complex relationship in a

data set

Previous time steps values were used as input and future

time steps values were obtained as output

Attempting to identify a pattern, and assuming that it will

continue in the future

Page 16: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

16

Artificial Neural Network(ANN) Technique (cont.) Four days’ data were used for training the network and

fifth day’s data were used for testing

During training, network takes the first 5 one minute

interval density value for computing the next one

Page 17: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

17

Model Based Approach

is the number of vehicles in the section at any instant of time

The time rate at which vehicles enter and exit the section respectively

The net time rate at which vehicles enter the section from the ramp

Aggregate speed of the vehicles in the section at any instant of time

Length of the section under study

Page 18: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

18

Model Based Approach (cont.) Discretizing the above equations using time step h

is cumulative number of vehicles in the section till kth

interval of time

Page 19: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

19

Outline Data Collection Prediction Techniques Comparison of The Techniques Conclusion

Page 20: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

20

Comparison of The Techniques (cont.)

is number of samples

is predicted number of vehicles

is measured number of vehicles

Mean Absolute Percentage Error

Page 21: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

21

Comparison of The Techniques (cont.)

Using model based technique

MAPE = 40.83%

Page 22: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

22

Comparison of The Techniques (cont.)

Using ANN based technique

MAPE = 32.47%

Page 23: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

23

Comparison of The Techniques (cont.)

Using historic technique

MAPE = 38.57%

Page 24: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

24

Outline Data Collection Prediction Techniques Comparison of The Techniques Conclusion

Page 25: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

25

Conclusion Implementing this application under a heterogeneous,

less lane disciplined traffic condition is more challenging.

Accuracy of the model based approach may improve if

collecting data for longer time period.

Accuracy of data driven techniques may improve if

collecting more video data.

Page 26: Prediction of Traffic Density for Congestion Analysis under Indian Traffic Conditions Proceedings of the 12th International IEEE Conference on Intelligent

26

Thank you for listening