Stability and Responsiveness of Real-Time Traffic Adaptive Systems

点击翻译按钮获取中文摘要

In this research, Real-time Traffic Adaptive Systems are modeled as Feedback Control Systems, where new decisions (the "phasing" in traffic network signal control and the ramp-metering rates in freeway control) are implemented based on "feedback" from sensors that monitor traffic. The feedback includes (i) a measurement system, (ii) a system that estimates/predicts traffic, and (iii) a decision/control system that computes appropriate decision/controls. In developing these systems, issues are raised on (i) the design of the feedback control system, (ii) the stability of the system, and (iii) the "responsiveness" of the system. The research will address these issues and particularly focus on: (1) Characterizing the communication/computational delays in the feedback (including the computational cycle of getting measurements, processing them, executing the embedded decision-making algorithms, processing the result, and downloading it to decision actuators /signals), (2) Characterizing the effect of design parameters in the feedback, such as "decision horizon", "control update interval", and "space-time resolution" of the decision-making algorithms, (3) Studying the stability of the system, in the sense does it go to an "equilibrium" or does it "cycle" in the limit, and (4) Investigating how well the feedback system responds to temporary short-term conditions (incidents, traffic surges, etc.) and slow varying long-term changes (e.g., gradual changes in the overall traffic volumes). In addition to basic research contributions and involvement of PhD students in the effort, it is anticipated that (1) the results on stability and responsiveness of traffic adaptive systems will be introduced into curricula, and (2) the analysis of the traffic adaptive systems would be very useful to USDOT and traffic agencies that are planning to implement state of-the-art systems within their jurisdictions.