Multivariate Optimization Strategies for Real-Time Traffic Control Signals

Abstract

The application of modern heuristic techniques, neural networks, simulated annealing, tabu search, and genetic algorithms for multivariate optimization is receiving increased attention compared with traditional techniques like hill climbing and gradient search. In the present research the efficiency and effectiveness of genetic algorithms are investigated for their application to the real-time optimization of traffic control signal timings and are compared with the efficiency and effectiveness of the Quasi-Newton gradient search method. The development, testing, comparison, and evaluation of these two multivariate optimization techniques for inclusion in the real-time traffic adaptive control system LOCAL model developed at the University of Pittsburgh as a part of an FHWA contract to a consortium led by the University of Maryland are described. The measures of effectiveness used in the comparisons include optimum total stopped delay, percentage of improvement in total stopped delay, optimal phase timings, execution time, and code size. Testing and evaluation results indicate that genetic algorithms are more efficient and effective than the Quasi-Newton method for this real-time application.

Get full access to this article

View all access options for this article.

References

Solanki

R. S.

, and Rathi

A. K.

. A Conceptual Framework for a Dynamic Traffic Flow Model. Proc., Advanced Traffic Management Conference on Large Urban Systems, St. Petersburg, Fla., FHWA, U.S. Department of Transportation, 1993.

Tarnoff

P. J.

, and Gartner

. Real-Time Traffic Adaptive Signal Control. Proc., Advanced Traffic Management Conference on Large Urban Systems, St. Petersburg, Fla., FHWA, U.S. Department of Transportation, 1993.

Dennis

J. E.

Jr. , and Schnabel

R. B.

. Unconstrained Optimization and Nonlinear Equations. Prentice-Hall, Englewood Cliffs, N.J., 1983.

Bunch

D. S.

A Comparison of Algorithms for Maximum Likelihood Estimation of Choice Models. Journal of Econometrics, Vol. 38, No. 1/2, May-June 1988, pp. 145–167.

Bullen

A. G. R.

, and Hummon

N. P.

. New Approaches to the Optimization and Simulation of a Vehicle-Actuated Traffic Signal. Proc., 1987 Engineering Foundation Conference on Management and Control of Urban Traffic Systems, Henniker, N. H., United Engineering Trustees, Inc., New York, 1988, pp. 91–102.

Y. C.

Traffic Flow Optimization on Rural Networks. Ph.D. dissertation. University of Maryland, College Park, 1983.

Davis

Handbook of Genetic Algorithms. Van Nostrand Reinhold, New York, 1991.

Goldberg

D. E.

Genetic and Evolutionary Algorithms Come of Age. Communications of the ACM, Vol. 37, No. 3, March 1994, pp. 113–119.

Janikow

C. Z.

Inductive Learning of Decision Rules from Attribute-Based Examples: A Knowledge-Intensive Genetic Algorithm Approach. Ph.D. dissertation. University of North Carolina, Chapel Hill, 1991.

10.

Wright

A. H.

Genetic Algorithms for Real Parameter Optimization. In Foundations of Genetic Algorithms, Morgan Kaufmann Publishers, San Mateo, Calif., 1991, pp. 205–218.