Traffic adaptive control for isolated, oversaturated intersections.

摘要

This dissertation presented the basic theory and evaluation of a new traffic adaptive control for over-saturated intersections or TACOS. TACOS is a hybrid optimization and rule-based adaptive control strategy. TACOS addresses at least in part problems of existing adaptive control strategies by having attributes such as: (1) The information used for decision-making processing is more reliable because it is not dependent on a forecasting model. (2) Intersection utilization is used explicitly as an objective function. (3) Phase sequencing is optimized. (4) Detection of operational anomalies and ability to respond to anomalies has been incorporated.; In the course of developing TACOS, three approaches for the estimation of queue length on each lane of intersection approaches were devised.; In order to evaluate TACOS, a new intersection simulator named NIT, was developed. NIT intersection simulator is a microscopic, stochastic, interval-oriented traffic simulator. It was designed to be able to simulate the intersection operation under pretimed, actuated and TACOS control. It can also be run with different car-following models, i.e., NETSIM, INTEGRATION, and NIT car following model. A great deal of performance comparisons of NIT, TSIS/NETSIM and INTEGRATION were conducted to validate NIT. Comprehensive test results showed that NIT with NIT car-following model was reliable to be used for TACOS evaluation.; For the purpose of evaluation, TACOS was compared with pretimed and actuated control strategies using NIT simulator with the NIT car-following model under increasingly complicated intersection geometry and flow scenarios. A comprehensive list of MOEs was collected and examined, including throughput-to-demand ratio, speed, delay, queue time, percent stops, and total green time per movement. TACOS showed significant improvements over pretimed and actuated control strategies in terms of all examined MOEs under all flow scenarios. In all cases, TACOS produced the best performance using the same parameter settings. However, the case studies indicated that when traffic loads are heavy, the effectiveness of TACOS could be damaged. The sensitive analysis of TACOS indicated that it is not particularly sensitive to key parameters. As a result, demanding parameter specifications which are required in other adaptive control strategies are avoided in TACOS. Overall, the simulation results indicated that TACOS is a promising improvement for adaptive traffic signal control.