Island Models for Stochastic Problem of Transportation Project Selection and Scheduling

The selection and scheduling of transportation projects can be expressed as a combinatorial optimization problem of finding the project implementation sequence that minimizes the total system cost over the analysis period. However, evaluating the total system cost is a challenging task for transportation researchers. Because of the uncertainty of travel times and project construction costs, the total system cost is usually stochastic instead of deterministic. This paper develops island models, which are variations of traditional genetic algorithms (GAs), for optimizing project selection and scheduling under resource constraints and explores the capability of the island models for solving a stochastic optimization problem. The total system cost is evaluated on the basis of equilibrium traffic assignment while a random term is introduced to emulate a stochastic environment. This work tests the sensitivity of the developed approach against the level of randomness and compares the results with those from traditional GAs. Analyses of results indicate that a well-designed island model is promising for solving a stochastic optimization problem.