Optimal Variable Speed Limit Control at a Lane Drop Bottleneck: Genetic Algorithm Approach

Abstract

This paper develops a genetic algorithm (GA) approach to solving the variable speed limit (VSL) control problem at a lane drop bottleneck. A multiobjective nonlinear integer model is formulated for the VSL control. The objective function includes the minimization of the sum of three components: the total travel time (TTT) on the studied freeway segments; the total speed variation (TSV) between the speed limits and the detected speeds from the most upstream and most downstream detectors; and the total speed difference (TSD) between the speed limits and the effective speeds on the controlled segments. Solution qualities from the GA and the sequential quadratic programming (SQP) algorithm are evaluated and compared. The numerical results show that the VSL control optimized by the GA outperforms the SQP. The VSL control results corresponding to various driver compliance rates are examined. The relationships among the truck percentages; the TTT, TSV, and TSD; and the combined objective function value are given. Finally, the potential effect of the left-lane truck restriction policy on the impact of trucks on the VSL control is examined and presented. The simulation results of the VSL control with left-lane truck restrictions slightly outperform those of the mixed traffic flow including cars and trucks.