http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
A Procedure for Determining Time-of-Day Break Points for Coordinated Actuated Traffic Signal Systems
Byungkyu(Brian) Park,이조영 대한토목학회 2008 KSCE JOURNAL OF CIVIL ENGINEERING Vol.12 No.1
This paper presents the development and evaluation of a procedure for determining optimal break points for time-of-day basedcoordinated actuated traffic signal operations. The proposed procedure uses a feature vector of optimal cycle length per time intervalinstead of traffic volume itself. Initial break points determined by the proposed feature vector are used in the greedy search algorithmthe number of evaluations in the search are dramatically reducedwhen compared to an exhaustive search or other common heuristic search methods such as a genetic algorithm. The proposedprocedure was evaluated using a hypothetical network consisting of four signalized intersections and the results indicated that theproposed procedure effectively improved the performance of the coordinated actuated signal control. In addition, sensitivity analysesresults on randomly varying demand conditions up to 20% and randomly increased demand conditions up to 30% indicated that thenewly developed break points were robust for such varying demand fluctuations.
Online Implementation of DynaMIT: A Prototype Traffic Examination and Prediction Program
Byungkyu(Brian) Park,이조영,Devi M. Pampati,Brian L. Smith 대한토목학회 2008 KSCE JOURNAL OF CIVIL ENGINEERING Vol.12 No.2
This paper presents a pilot study of conducting an online implementation of DynaMIT, one of traffic estimation and prediction(TrEP) programs developed by an MIT research team with the support from the U.S. Federal Highway Administration, in HamptonRoads, VA, U.S.A. As a first step, a test-bed network was cbefore evaluating the performance of the online implementation of DynaMIT. Based on the online implementation of DynaMIT forthree days, it was found that DynaMIT showed fairly good performance on the estimation and prediction of the sensor counts withthe root mean square normalized (RMSN) error ranges betwen 0.15 and 0.25 for estimations, and 0.25 and 0.4 for predictions. Eventhough speed and travel times showed some discrepancies, further investigations indicated that the performance of DynaMIT can besignificantly improved with adequately calibrated supply parameters.
Byungkyu "Brian" Park,Kangyuan Zhu 대한토목학회 2007 KSCE journal of civil engineering Vol.11 No.4
An estimation of origin-destination (OD) demand matrix is one of key elements to ensure the success in traffic modeling analysis. Even with widely deployed traffic sensors and advanced computational technologies, an estimation of time-dependent OD matrices is still a key barrier for the implementation of dynamic traffic assignment as well as simulation-based traffic modeling analysis. This paper proposes an improvement of existing time-dependent OD estimation method by updating assignment matrix at each step of OD estimation process and quantifies benefits and costs of doing so. The results from a case study with Florian network showed that the estimated OD flows from the proposed GA-based method with updated assignment matrix reduced the sum of square errors by 40% when compared with the OD flows from the DynaMIT OD estimation method with fixed assignment matrix, one of the most commonly used OD estimation methods. However, the proposed method would require significantly higher computational time than the traditional DyanMlT OD estimation method.
Comparison of Emergency Vehicle Preemption Methods Using a Hardware-in-the-Loop Simulation
Yun, Ilsoo,Park, Byungkyu Brian,Lee, Choul Ki,Oh, Young Tae 대한토목학회 2012 KSCE JOURNAL OF CIVIL ENGINEERING Vol.16 No.6
Modern traffic signal control systems provide Emergency Vehicle Preemption (EVP), which provides the right-of-way to emergency vehicles in preference to other vehicles. Traffic engineers should select the best coordination recovery strategy at the end of the preemption in order to minimize any resulting disruptions. This paper compares the various EVP methods available with a controller used in the USA. The comparison was conducted in an urban corridor including four coordinated-actuated signals using a Hardware-In-the-Loop Simulation (HILS), which consisted of a well-calibrated VISSIM microscopic simulation model and four Type 170 controllers. The examined preemption strategies included shortway and dwell with different numbers of cycles for transition completion. The results of the comparison indicated that the resulting impact on the network-wide and intersection-specific performances varied with the selected EVP method. The results also indicated that the EVP function using a shortway transition method with two or three cycles worked the best, as it could minimize the impacts of EVP.
Preceding Vehicle Identification for Cooperative Adaptive Cruise Control Platoon Forming
Chen, Zheng,Park, Byungkyu Brian IEEE 2020 IEEE transactions on intelligent transportation sy Vol.21 No.1
<P>Cooperative adaptive cruise control (CACC) has shown great potential in enhancing traffic efficiency and sustainability. While past research efforts focused on the development of CACC systems and their demonstrations, very few of them considered in detail how to form a CACC platoon in real traffic, where the proper identification of preceding vehicle is required. To ensure safe and reliable CACC operations, the following vehicle needs to establish the correct connection with its preceding vehicle. Although this can be done by matching information shared by surrounding vehicles with the ego-vehicle’s radar measurements, the existence of sensor/global positioning system (GPS) errors makes it a challenging task. Considering possible sensor/GPS errors in real traffic, this paper proposes a procedure of identifying preceding vehicle under fully connected vehicle environment and evaluates three preceding vehicle identification systems (PVIS), namely, location-based PVIS, distance-based PVIS, and integrated PVIS combining both location and distance information. The mathematical models of PVISs are developed. The performance evaluation of the PVISs is conducted based on real vehicle trajectory data from the Next Generation Simulation (NGSIM) program, which reflects how vehicles’ relative positions change in a high-density segment of highway. The feasibility, performance, and potential of the three PVISs are compared. The results show that location-based PVIS requires a relative positioning accuracy below 1.1 m to ensure an acceptable identification time with zero failure rate. The integrated PVIS has the best performance, providing 99% confidence in identifying preceding vehicle within 1.3 s under typical sensor error settings.</P>
Agbolosu-Amison, Seli James,Yun, Ilsoo,Park, Byungkyu Brian 대한토목학회 2012 KSCE Journal of Civil Engineering Vol.16 No.3
Traffic signal timing optimization and control are one of the most cost-effective ways of improving urban arterial network congestions. Among various control modes, actuated traffic signal control is designed to provide green times where they are needed and it uses a pre-specified gap-out time to determine early termination of current phase green time. However, the effectiveness of its signal state decisions is limited by its dependence on vehicular information from fixed point sensors. With the emerging wireless communication technology based on cooperative vehicle-infrastructure system known as IntelliDrive, individual vehicular information (e.g., speed, acceleration, and location) is needed to be fully utilized for traffic signal control applications. This paper presents a dynamic gap-out approach that uses individual vehicular information and quantified benefits of the proposed approach via simulation. The timing plans were optimized by the genetic algorithm for both the traditional and the proposed gap-out cases. The results based on a four-leg intersection indicated that the dynamic gap-out reduced vehicular delays by 12.5% over existing regular gap-out.