Author

Faisal Ahmed

Date of Award

Summer 5-2014

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Civil Engineering

First Advisor

Dr. Yaser E. Hawas

Second Advisor

Dr. Kamran Ahmed

Third Advisor

Dr. Younes Hamdouch

Abstract

The aim of this thesis is to develop a distributed adaptive control system which can work standalone for a single intersection to handle various boundary conditions of recurrent, non-recurrent congestion, transit signal priority and downstream blockage to improve the overall network in terms of productivity and efficiency. The control system uses link detectors’ data to determine the boundary conditions of all incoming and exit links. Four processes or modules are deployed. The traffic regime state module estimates the congestion status of the link. The incident status module determines the likelihood of an incident on the link. The transit priority module estimates if the link is flagged for transit priority based on the transit vehicle location and type. Finally, the downstream blockage module scans all downstream links and determines their recurrent blockage conditions. Three different urban incident detection models (General Regression Model, Neuro-Fuzzy Model and Binary Logit Model) were developed in order to be adopted for the incident status module. Among these, the Binary Logit Model was selected and integrated with the signal control logic. The developed Binary Logit Model is relatively stable and performs effectively under various traffic conditions, as compared to other algorithms reported in the literature. The developed signal control logic has been interfaced with CORSIM micro-simulation for rigorous evaluations with different types of signal phase settings. The proposed system operates in a manner similar to a typical pre-timed signal (with split or protected phase settings) or a fully actuated signal (with splitphase arrangement, protected phase, or dual ring phase settings). The control decisions of this developed control logic produced significant enhancement to productivity (in terms of Person Trips and Vehicle Trips) compared with the existing signal control systems in medium to heavily congested traffic demand conditions for different types of networks. Also, more efficient outcomes (in terms of Average Trip Time/Person and delay in seconds/vehicle) is achieved for relatively low to heavy traffic demand conditions with this control logic (using Split Pre-timed). The newly developed signal control logic yields greater productivity than the existing signal control systems in a typical congested urban network or closely spaced intersections, where traffic demand could be similarly high on both sides at peak periods. It is promising to see how well this signal control logic performs in a network with a high number of junctions. Such performance was rarely reported in the existing literature. The best performing phase settings of the newly developed signal control were thoroughly investigated. The signal control logic has also been extended with the logic of pre-timed styled signal phase settings for the possibility of enhancing productivity in heavily congested scenarios under a closely spaced urban network. The performance of the developed pre-timed signal control signal is quite impressive. The activation of the incident status module under the signal control logic yields an acceptable performance in most of the experimental cases, yet the control logic itself works better without the incident status module with the Split Pre-timed and Dual Actuated phase settings. The Protected Pre-timed phase setting exhibits benefits by activating the incident status module in some medium congested demands

Share

COinS