Sei sulla pagina 1di 2

VISGAOST - VISSIM-BASED GENETIC ALGORITHM OPTIMIZATION OF SIGNAL TIMINGS

Aleksandar Stevanovic, Peter T. Martin, Jelka Stevanovic - Utah Traffic


Lab, Department of Civil & Environmental Engineering, University of Utah

VISGAOST - Program which uses a Genetic Algorithm to optimize signal timings based on their
evaluations in VISSIM. VISGAOST is the first optimization tool which can optimize controller’s Transit
Signal Priority (TSP) settings. VISGAOST can optimize basic signal timing parameters (i.e. cycle length,
green splits, offsets and phase sequences) simultaneously (or separately) with the TSP settings. Virtually any
VISSIM performance measure can be used as objective function. For example, VISGAOST can optimize
signal timings to reduce pollutant emissions reported by VISSIM ( currently only optimization software which
can optimize signal timings to reduce traffic emissions in the network). The list of special VISSIM features
also include the optimization of: phasing sequence, multiple coordinated systems and uncoordinated
intersections, fully-actuated isolated intersections, and multiple time periods.

Basic Algorithm VISGAOST Dialog Windows


Initialize First Population

Encode First Population

Simulate & Evaluate Population

Create Next Generation of Population


Ÿ Elitism
End Criteria Ÿ Crossover
no
Satisfied? Ÿ Mutation
Ÿ Population Simulation and
Evaluation

yes

Return Best Solution

Binary Encoding
Before encoding After encoding
5 4 3 2 1 0
Parameter value 13 = 0*2 + 0*2 + 1*2 + 1*2 + 0*2 + 1*2 001101

Roulette Wheel Selection


Fitness value Selection probability
5% 15% Chromosome1 40 50%
50% Chromosome2
30% 24 30%
Chromosome3
12 15%
Chromosome4
4 5%
Total
80 100%

Crossover
Offspring
Parent1 Parent2
1 1 0 1 0 0 0
1 1 0 1 0 1 1 + 0 0 0 0 1 0 0 =
0 0 0 0 1 1 1
Crossover point
Mutation
After crossover After mutation
1 1 0 1 0 1 1 1 0 0 1 1 1 1

Inverted bits
0
VISSIM Model of Park City, UT
0.5 1 mile
Results - 100 Simulation Runs
Scenario 1: 15-min Field SYNCHRO VISGAOST

peak intersection

traffic Mean St. Dev Mean St. Dev Mean St. Dev

Performance Index 75.63* 5.39 71.48^ 6.17 63.3^* 3.9

Delay (hours) 54.52* 3.11 49.17^ 3.44 46.78^* 2.19

Stops 7600* 869 8031^ 1164 5946 ^* 662

Throughput (veh) 2183* 28 2200^ 26 2219 ^* 28

Scenario 2: 15-min Field SYNCHRO VISGAOST

peak network traffic Mean St. Dev Mean St. Dev Mean St. Dev

Performance Index 73.38* 5.32 70.67^ 8.64 59.07^* 3.66

Optimization of Signal Timings Delay (hours) 53.09* 3.13 48.82^ 5.37 43.54^* 2.17

Stops 7303* 838 7864^ 1667 5589 ^* 582


130
Average GA

120 Best SYNCHRO Throughput (veh) 2204* 26 2214^ 49 2257^* 29


110 Field Timings

Best GA
Scenario 3: 1-hour Field SYNCHRO VISGAOST
100
Performance Index

peak network traffic Mean St. Dev Mean St. Dev Mean St. Dev
90

80 Performance Index 271.4* 16.8 299^ 55.14 227.46^* 12.06


70
Delay (hours) 194.21 12.35 193.47 25.58 164.47^* 6.27
60

All Timings
Cycle Lengths
&
Offsets
&
Splits Stops 27792* 2590 37985^ 1 2173 22677^* 2157
50
Offsets Phase Sequence

40 Throughput (veh) 8556 73 8553 101 8599^* 38


0 400 800 1200 1600 2000 2400 2800 3200 3600 4000 4400 4800 5200 5600 6000 6400 6800 7200 7600 8000 8400 8800 9200
Number of evaluations

Central Ave. Albany, NY in VISSIM Optimization of TSP &


Basic Signal Timings
15.5

15.25 Initial Signal Timings

Optimization of TSP Timings


15
Optimization of Basic Signal Timings
Average Person Delay (sec)

14.75 Optimization of All Signal Timings

14.5

14.25

14

13.75

13.5
1 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199
Generations