A Systematic Genetic Algorithm Based Framework to Optimize Intelligent Transportation System (ITS) Strategies

摘要

Today more and more Intelligent Transportation System (ITS) strategies such as High Occupancy Toll (HOT), ramp metering, variable speed limits, etc., are introduced nationwide to reduce congestion and maintain desired service levels on the freeway. Optimization of all the different parameters in such strategies is vital to manage the traffic effectively. The paper introduces a modeling and optimization framework to calibrate and optimize multiple ITS strategies simultaneously. The framework is capable of estimating the OD matrix, calibrating the model with the experimental data, and conducting system optimization in transportation. The approach can be used to optimize more complex system optimization problems in transportation domain. In this paper, this capability is shown with simulation experiments based on the public data of Ⅰ-95 Florida. The preliminary results show that the revenue increases with target speed up to a point, after which the revenue drops; the throughput on the HOT lanes increases with decreasing target speed. So, to maximize revenue and throughput together, the optimal parameters exist in the middle of the feasible range. Additionally, when accidents happen, the total throughput can be improved by decreasing the toll rate.