01595111

Component

Microsimulation modeling is one of the contemporary techniques that provides an environment to experiment and test new techniques without any disruption to the traffic in a real network and without having the new technology actually installed on field. For accurate and reliable results, the simulation models need to be well calibrated and validated. Most traffic simulation software has many user-adjustable parameters that are attuned for calibrating the model. Manual calibration is time consuming and is often no better than a trial and error approach. Researchers have used different algorithms for calibrating simulation models. The results from past performance comparisons are mixed indicating that the choice of best heuristics for a given problem depends on the nature of problem. The use of Genetic Algorithm (GA) and Simulated Annealing (SA) for calibrating microsimulation traffic models have been utilized in past studies independently without evaluating which performs better. This paper therefore addresses the need to examine which of the two (GA and SA) heuristics perform better for calibrating microsimulation models. The same microsimulation model is calibrated using these two heuristic algorithms and their performances are evaluated. The results show that the heuristics approach can be resorted to for calibrating simulation models very effectively as it offers the benefit of automating the cumbersome calibrating process. Both GA and SA have the ability to find better calibrating parameters than the manually calibrated parameters. The number of better solutions as well as the best solution found by SA are better than those found by GA. However, the best solution found by GA is very close to that of SA. Thus, it can be concluded that SA and GA equally perform better for calibrating the microsimulation models. The approach presented in this research can be used to help engineers and planners achieve better modeled results.

This paper was sponsored by TRB committee AHB45 Standing Committee on Traffic Flow Theory and Characteristics. Alternate title: Comparative Study of Simulated Annealing and Genetic Algorithm for Calibration of VISSIM Microsimulation Model

01584066

16-4211

English

23p

2016

Transportation Research Board 95th Annual Meeting

Digital/other

Appendices; Figures; Maps; References; Tables

Genetic algorithms; Microsimulation; Simulated annealing; Traffic control; Traffic simulation; Transportation planning

VISSIM (Computer model)

Highways; Operations and Traffic Management

Transportation Research Board Annual Meeting 2016 Paper #16-4211

TRIS, TRB, ATRI

Jan 12 2016 5:51PM