Jizhan Gou


Zhang, Li

Committee Member

White, Thomas D.

Committee Member

McAnally, William H.

Committee Member

Jin, Mingzhou

Date of Degree


Document Type

Dissertation - Open Access


This dissertation proposes a Dynamic Message Signs (DMS) diversion control system based on principles of existing Advanced Traveler Information Systems and Advanced Traffic Management Systems (ATMS). The objective of the proposed system is to alleviate total corridor traffic delay by choosing optimized diversion rate and alternative road signal-timing plan. The DMS displays adaptive messages at predefined time interval for guiding certain number of drivers to alternative roads. Messages to be displayed on the DMS are chosen by an on-line optimization model that minimizes corridor traffic delay. The expected diversion rate is assumed following a distribution. An optimization model that considers three traffic delay components: mainline travel delay, alternative road signal control delay, and the travel time difference between the mainline and alternative roads is constructed. Signal timing parameters of alternative road intersections and DMS message level are the decision variables; speeds, flow rates, and other corridor traffic data from detectors serve as inputs of the model. Traffic simulation software, CORSIM, served as a developmental environment and test bed for evaluating the proposed system. MATLAB optimization toolboxes have been applied to solve the proposed model. A CORSIM Run-Time-Extension (RTE) has been developed to exchange data between CORSIM and the adopted MATLAB optimization algorithms (Genetic Algorithm, Pattern Search in direct search toolbox, and Sequential Quadratic Programming). Among the three candidate algorithms, the Sequential Quadratic Programming showed the fastest execution speed and yielded the smallest total delays for numerical examples. TRANSYT-7F, the most credible traffic signal optimization software has been used as a benchmark to verify the proposed model. The total corridor delays obtained from CORSIM with the SQP solutions show average reductions of 8.97%, 14.09%, and 13.09% for heavy, moderate and light traffic congestion levels respectively when compared with TRANSYT-7F optimization results. The maximum model execution time at each MATLAB call is fewer than two minutes, which implies that the system is capable of real world implementation with a DMS message and signal update interval of two minutes.