Travel Time Estimation with Correlation Analysis of Single Loop Detector Data

Abstract

In this paper, the average travel time over a link is considered as a random variable following an identical probability distribution as the arrival process. A new estimation method of the average travel time uses a cross-correlation analysis of traffic flow measurement data. This method requires only traffic flow information, which is available from the measurements of single loop detectors upstream and downstream from one link. Different from the existing maximum cross-correlation analysis method, the proposed method considers average travel time as a random variable, with its mean value estimated from all significant cross-correlation coefficients rather than from only the maximum cross-correlation coefficient. Therefore, the inherent variability of average travel time among different vehicles can be considered. Moreover, different from the existing optimization method, the proposed method uses the statistical t-test of the significant cross-correlation coefficients to determine automatically and adaptively the fitting range of the probability density function of the average travel time. Thus, it avoids using the approximated car length factor and has no need to predetermine the range of the average travel time as required by the optimization method. Details of the average travel time estimation procedures are presented, and the effectiveness of the proposed method is demonstrated through both simulation study and a case study of real traffic data.

Automatic Incident Detection at Intersections with Use of Telematics

“…An incident should extend the time in which the vehicles cover the distance between the upstream and downstream detector due to the obstruction, which implies an incident. The speed of a vehicle can be detected using single inductive loop at junction entries and exits, correlation analysis of traffic volume on the inductive loop (Guo, 2006) or advanced algorithms that allows the re-identification of the vehicle on the junction by using its electric signature from the inductive loop (Sun, 2004). The average value for the total number of vehicles passing through the junction can be calculated by treating all upstream detectors as one entry detector and all downstream detectors as one exit detector.…”

Produção de dados de tráfego sintéticos através de algoritmo genético e simulação microscópica

“…Essas técnicas geram dados sintéticos, que podem ser velocidades ou tempos de viagem (Petty et al, 1998;Lin et al, 2004;Jeong e Rilett, 2005;Guo e Jin, 2006;Oh e Chung, 2006;Emam e Al-Deek, 2006) ou ainda matrizes OD (Gajewski et al, 2002;Park et al, 2008). Diversos estudos usam dados gerados por simulação para complementar dados coletados em campo (Schultz, 2003;Gomes, 2004) ou até para substituí-los (Araújo e Setti, 2008).…”

Hybrid Approaches to Estimating Freeway Travel Times Using Point Traffic Detector Data

“…The methods range from simple regression analyses and cross correlation analyses (Dailey 2003), to more complicated probabilistic regression models (Petty et al 1998;Zhang et al 1999;Guo and Jin 2006). developed two regression models to estimate travel time: one for freeways with a permanent bottleneck, and another for freeways with a temporary downstream bottleneck.…”

“…The case study showed that this method can estimate travel time even during very congested conditions. Later, extensions of this method include using the B-splines, exponential moving (Zhang et al 1999), and correlation analysis (Guo and Jin 2006) to improve the estimation of the probability density function.…”

“…In addition, using traffic volumes and occupancies to estimate speeds may not be appropriate for uncongested conditions since the speed is not sensitive to traffic demand under these conditions (Vanajakshi 2004). Statistics-based methods such as the probabilistic regression are proven to be robust during congested conditions (Guo 2006). However, such methods usually use a data sampling interval of 1-second, which is not commonly available in practice, where traffic detector data are commonly polled at 20-second or 30-second frequencies in current traffic management applications.…”