Tabu Search Strategies for Variable Speed Limit Control at a Lane Drop Bottleneck

Abstract

A tabu search–based algorithm is developed in this study and applied for optimizing a variable speed limit (VSL) control system. A macroscopic traffic flow model is used to predict the traffic states over the prediction horizon. The objective of the VSL control is to minimize the total value of travel time and total value of speed variation on the selected freeway segments. Sensitivity analyses are conducted for the tabu search parameters. Different weight sets of the objective function are selected and tested. Solution qualities from the tabu search algorithm and sequential quadratic programming (SQP) algorithm are compared. Numerical results clearly indicate that the proposed tabu search outperforms the SQP, which is used as a benchmark. The control results also show that the VSL can effectively reduce total travel time and total speed variation at a lane drop bottleneck. The relationships between the number of VSL control segments and total travel time, total speed variation, and combined objective function value corresponding to different weight sets are also given.