Microscopic Traffic Flow Models Research Articles

Empirical test of a microscopic three-phase traffic theory

A review of dynamic nonlinear features of spatiotemporal congested patterns in freeway traffic is presented. The basis of the review is a comparison of theoretical features of the congested patterns that are shown by a microscopic traffic flow model in the context of the Kerner's three-phase traffic theory and empirical microscopic and macroscopic pattern characteristics measured on different freeways over various days and years. In this test of the microscopic three-phase traffic flow theory, a model of an "open" road is applied: Empirical time-dependence of traffic demand and drivers' destinations are used at the upstream model boundaries. At downstream model boundary conditions for vehicle freely leaving a modeling freeway section(s) are given. Spatiotemporal congested patterns emerge, develop, and dissolve in this open freeway model with the same types of bottlenecks as those in empirical observations. It is found that microscopic three-phase traffic models can explain all known macroscopic and microscopic empirical congested pattern features (e.g., probabilistic breakdown phenomenon as a first-order phase transition from free flow to synchronized flow, moving jam emergence in synchronized flow rather than in free flow, spatiotemporal features of synchronized flow and general congested patterns at freeway bottlenecks, intensification of downstream congestion due to upstream congestion at adjacent bottlenecks). It turns out that microscopic optimal velocity (OV) functions and time headway distributions are not necessarily qualitatively different, even if local congested traffic behavior is qualitatively different. Model performance with respect to spatiotemporal pattern emergence and evolution cannot be tested using these traffic characteristics. The reason for this is that important spatiotemporal features of congested traffic patterns are lost in these and many other macroscopic and microscopic traffic characteristics, which are widely used as the empirical basis for a test of traffic flow models.

Models of vehicular traffic: An engineering perspective

The aim of the paper is to present an engineer's viewpoint of traffic streams and their models both at the macroscopic and microscopic levels. The paper concentrates on two classes of macroscopic models (namely, stream description models and travel time estimation models). At the microscopic level the paper concentrates on car-following models and also presents a relatively recent idea on developing a comprehensive microscopic model of driver behaviour. Finally, the paper presents some properties which all microscopic models of traffic flow should possess and also tries to identify areas where research will bring about a qualitative jump in the understanding of how traffic flows.

Deterministic microscopic three-phase traffic flow models

Two different deterministic microscopic traffic flow models, which are in the context of the Kerner's there-phase traffic theory, are introduced. In an acceleration time delay model (ATD-model), different time delays in driver acceleration associated with driver behaviour in various local driving situations are explicitly incorporated into the model. Vehicle acceleration depends on local traffic situation, i.e., whether a driver is within the free flow, or synchronized flow, or else wide moving jam traffic phase. In a speed adaptation model (SA-model), vehicle speed adaptation occurs in synchronized flow depending on driving conditions. It is found that the ATD- and SA-models show spatiotemporal congested traffic patterns that are adequate with empirical results. In the ATD- and SA-models, the onset of congestion in free flow at a freeway bottleneck is associated with a first-order phase transition from free flow to synchronized flow; moving jams emerge spontaneously in synchronized flow only. Differences between the ATD- and SA-models are studied. A comparison of the ATD- and SA-models with stochastic models in the context of three phase traffic theory is made. A critical discussion of earlier traffic flow theories and models based on the fundamental diagram approach is presented.

Probabilistic breakdown phenomenon at on-ramp bottlenecks in three-phase traffic theory: Congestion nucleation in spatially non-homogeneous traffic

A nucleation model for the breakdown phenomenon in an initial non-homogeneous free traffic flow that occurs at an on-ramp bottleneck is presented. This model is in the context of three-phase traffic theory. In this theory, the breakdown phenomenon is associated with a first-order phase transition from the “free flow” phase to the “synchronized flow” phase. In contrast with many other nucleation models for phase transitions in different systems of statistical physics in which random precluster emergence from fluctuations in an initial homogeneous system foregoes subsequent cluster evolution towards a critical cluster (critical nuclei), random precluster occurrence in free flow at the bottleneck is not necessary for traffic breakdown. In the model, the breakdown phenomenon can also occur if there were no fluctuations in free flow. This is because there is a permanent and motionless non-homogeneity that can be considered a deterministic vehicle cluster localized in a neighborhood of the bottleneck. The presented nucleation model and a nucleation rate of traffic breakdown that follows from the model exhibit qualitatively different features in comparison with previous results. In the nucleation model, traffic breakdown nucleation occurs through a random increase in vehicle number within the deterministic vehicle cluster, if the amplitude of the resulting random vehicle cluster exceeds some critical amplitude. The mean time delay and the associated nucleation rate of traffic breakdown at the bottleneck are found and investigated. The nucleation rate of traffic breakdown as a function of the flow rates to the on-ramp and upstream of the bottleneck is studied. Boundaries for traffic breakdown in the diagram of congested patterns at the bottleneck are found. These boundaries are qualitatively correlated with numerical results of simulation of microscopic traffic flow models in the context of three-phase traffic theory.

Control of spatiotemporal congested traffic patterns at highway bottlenecks

The physics of congested traffic pattern formation at an on-ramp bottleneck under on-ramp metering is studied. Various congested pattern nucleation effects at the bottleneck under different control rules are found. Based on the author's three-phase traffic theory, a congested pattern feedback on-ramp inflow control approach is proposed and investigated. In this congested pattern control approach (ANCONA approach for short), congestion at an on-ramp bottleneck is allowed to set in. The basis idea of the ANCONA approach is to keep congestion conditions at the minimum possible level at the bottleneck. In particular, the congested pattern should not propagate upstream, i.e., the congested pattern should be localized on the main road in a small neighborhood of the bottleneck. It is used in the ANCONA approach that in accordance with empirical congested pattern features, congested patterns are of various types some more favorable than others in terms of the discharge volume from the congested pattern and of the vehicle delay time due to congestion. Based on a microscopic traffic flow model of Kerner and Klenov in the context of three-phase traffic theory, a comparison of the ANCONA approach with a well-known free flow control approach in which free flow is maintained at the bottleneck is made. In particular, a ramp metering ALINEA method based on the free flow control approach that is used in many real installations is studied and compared with the ANCONA approach. Benefits of the ANCONA approach are (i) higher throughputs on the main road downstream of the bottleneck, (ii) considerably lower vehicle waiting times at the light signal on the on-ramp, (ii) the upstream propagation of congestion does not occur even if large amplitude perturbations appear in traffic flow.

Calibration and Validation of Microscopic Models of Traffic Flow

Because microscopic models are heavily used in applications, the appropriate calibration and validation of these models have been a recent concern. The contribution of this paper is to compare some of these models by calibrating and validating them with data from double loop detectors on a multilane freeway. To simplify this task, the models were tested by simplifying the multilane reality to a simulation of only a single lane. The results show that by simulating the multilane road with single-lane models, calibration errors (Theil's U-value, or the root mean square error) of 14% to 16% were obtained. A validation of the models, which was done by taking the calibrated parameters of one data set to reproduce the other data sets, gives additional errors of about 0.5 to 2.5 percentage points. This is in good agreement with other calibration and validation approaches performed recently.

Calibration and Validation of Microscopic Models of Traffic Flow

Because microscopic models are heavily used in applications, the appropriate calibration and validation of these models have been a recent concern. The contribution of this paper is to compare some of these models by calibrating and validating them with data from double loop detectors on a multilane freeway. To simplify this task, the models were tested by simplifying the multilane reality to a simulation of only a single lane. The results show that by simulating the multilane road with single-lane models, calibration errors (Theil's U-value, or the root mean square error) of 14% to 16% were obtained. A validation of the models, which was done by taking the calibrated parameters of one data set to reproduce the other data sets, gives additional errors of about 0.5 to 2.5 percentage points. This is in good agreement with other calibration and validation approaches performed recently.

Analysis and Comparison of Microscopic Traffic Flow Models with Real Traffic Microscopic Data

The evermore widespread use of microscopic traffic simulation in the analysis of road systems has refocused attention on submodels, including car-following models. The difficulties of microscopic-level simulation models in the accurate reproduction of real traffic phenomena stem not only from the complexity of calibration and validation operations but also from the structural inadequacies of the submodels themselves. Both of these drawbacks originate from the scant information available on real phenomena because of the difficulty with the gathering of accurate field data. In this study, the use of kinematic differential Global Positioning System instruments allowed the trajectories of four vehicles in a platoon to be accurately monitored under real traffic conditions on both urban and extraurban roads. Some of these data were used to analyze the behaviors of four microscopic traffic flow models that differed greatly in both approach and complexity. The effect of the choice of performance measures on the model calibration results was first investigated, and intervehicle spacing was shown to be the most reliable measure. Model calibrations showed results similar to those obtained in other studies that used test track data. Instead, validations resulted in higher deviations compared with those from previous studies (with peaks in cross validations between urban and extraurban experiments). This confirms the need for real traffic data. On comparison of the models, all models showed similar performances (i.e., similar deviations in validation). Surprisingly, however, the simplest model performed on average better than the others, but the most complex one was the most robust, never reaching particularly high deviations.

Analysis and Comparison of Microscopic Traffic Flow Models with Real Traffic Microscopic Data

The evermore widespread use of microscopic traffic simulation in the analysis of road systems has refocused attention on submodels, including car-following models. The difficulties of microscopic-level simulation models in the accurate reproduction of real traffic phenomena stem not only from the complexity of calibration and validation operations but also from the structural inadequacies of the submodels themselves. Both of these drawbacks originate from the scant information available on real phenomena because of the difficulty with the gathering of accurate field data. In this study, the use of kinematic differential Global Positioning System instruments allowed the trajectories of four vehicles in a platoon to be accurately monitored under real traffic conditions on both urban and extraurban roads. Some of these data were used to analyze the behaviors of four microscopic traffic flow models that differed greatly in both approach and complexity. The effect of the choice of performance measures on the m...

Performance Evaluation of Microscopic Traffic Flow Models with Test Track Data

The performance of six microscopic traffic flow models was investigated on the basis of how well these models fit with the real-time kinematic Global Positioning System (GPS) measurements. Ten passenger cars equipped with the GPS receivers participated in the car-following experiments, conducted at a test track. The genetic algorithm-based approach is adopted to optimize the model parameters for two different cases: using speed and headway data. The optimized performance of each model is analyzed for various driving conditions introduced by the different level of disturbances to the lead vehicle's speed, which include half-wave, one-wave, two-wave, three-wave, random, and constant speed patterns. In the former case with speed data, five models performed well with the average percentile error ranging from 3.87% to 4.71% and standard deviation ranging from 1.09% to 1.64%. In the latter case with headway data, only three models performed well with the average percentile error ranging from 12.04% to 12.91% and standard deviation ranging from 4.53% to 5.13%. All models performed better in the former case than in the latter case. The interpersonal variations are significant compared with the intermodel variations and indicate individual drivers' influence on the car-following phenomena.

Calibration and Validation of Microscopic Traffic Flow Models

Microscopic simulation models are becoming increasingly important tools in modeling transport systems. A large number of models are used in many countries. The most difficult stage in the development and use of such models is the calibration and validation of the microscopic submodels such as the car-following and gap acceptance models. This difficulty results from the lack of suitable methods for adapting models to empirical data. Recent progress has been made in calibrating a number of microscopic traffic flow models. Ten very different models have been tested with data collected via Differential Global Positioning System-equipped cars on a test track in Japan. To calibrate the models, the data of the leading car are fed into the model under consideration, and the model is used to compute the headway time series of the following car. The deviations between the measured and the simulated headways are then used to calibrate and validate the models. The calibration results agree with earlier studies as there are errors of 12% to 17% for all models, and no model can be denoted to be the best. The differences between individual drivers are larger than the differences between different models. The validation process gives acceptable errors from 17% to 22%. But for special data sets with validation errors up to 60%, the calibration process has reached what is known as "overfitting": because of the adaptation to a particular situation, the models are not capable of generalizing to other situations.

Toward Benchmarking of Microscopic Traffic Flow Models

Several microscopic traffic flow models were tested with a publicly available data set. The task was to predict the travel times between several observers along a one-lane rural road, given as boundary conditions the flow into this road and the flow out of it. By using nonlinear optimization, the best matching set of parameters for each of the models was estimated. For this particular data set, the models that performed best were the ones with the smallest number of parameters. The average error rate of the best models is about 16%; however, this value is not very reliable: the error rate fluctuates between 2.5% and 25% for different parts of the data set.

Analytic, Microscopic Model of Traffic Flow and Travel Time

A model of disaggregate interactions between individual vehicles was developed that makes explicit the vehicle classes, the class trip rates, and their kinematic parameters (free speed, acceleration rate, length, safety margin). Assuming exponential gaps between vehicles, analytic formulas were derived for the mean value and the variance of the path travel times within each class. The model was successively applied to one-lane roads; two-way, two-lane roads; one-way, two-lane roads; and two-way, three-lane roads.

Design and Implementation of Control-Theory-Based Microscopic Traffic Flow Model

A microscopic traffic flow model based on the constant-time-headway policy and McRuer’s man-machine crossover model was designed. Automatic control theory concepts were employed in the model formulation. The constant-time-headway policy was used to generate the command model of a human driver’s decision for vehicle acceleration or deceleration. This command is the input signal fed into the driver-vehicle dynamics suggested by the crossover model. The proposed model was mathematically formulated, designed, implemented, and numerically simulated. The stability properties and validity of the proposed model were analyzed on the basis of the simulation results. It was demonstrated that the proposed model can reproduce well-known traffic phenomena such as shock waves, intersection starting and stopping waves, and loop structures of flow-density and speed-density plots.

Development and Performance Evaluation of an ITS-Ready Microscopic Traffic Model for Irvine, California

The research in this paper presents the detailed development and on-line evaluation of a microscopic traffic flow model for the city of Irvine in Southern California. This effort is the first stage of evaluating micro-simulators in terms of their ability to model and analyze Intelligent Transportation Systems (ITS) under faster-than-real-time conditions. We utilize "Paramics," a particularly promising ITS-capable advanced traffic flow simulator and visualization tool, as one of an array of newly emerging ITS-capable simulation tools and we apply it to the Irvine network as part of a staged effort to model the much larger Southern California network. The driver and vehicle models and parameters were developed to reflect U.K. driver and vehicle characteristics. In this effort we explain our procedure used the calibrate these parameters to reproduce local U.S. traffic behavior. We built a model of a conventional U.S. freeway/arterial network in Southern California and calibrated its parameters using on-line field data. The calibrated models are validated, both at the section and network levels, and evaluated relative to their potential application in Advanced Traffic Management and Information Systems (ATMIS). Based on obtained results, the calibrated model performed well during validation on a freeway link. On the full network, the vehicle release mechanism showed some time-lag in releasing demand onto the network. This is potentially due to stacking of vehicles in memory before adequate headways are found on the road to release the vehicles. Although the problem itself is simple, its effects on the results were notable.

Asymmetric Optimal Velocity Model for Traffic Flow

A lot of attempts for constructing models for traffic phenomena have been developed in the view points of physics in recent several years. The car-following model, which has a long history of early works for studying traffic flow, is a kind of models for a kind of microscopic traffic flow models. The Optimal Velocity (OV) model has first succeeded in showing the spontaneous jam-formation in this type of models as the phenomena of phase transition 1) and well reproduced the real data in the fundamental diagram, which show the relation between flow and car-density 2) and the data of the car-following behavior. The model is formulated as

Parametrization of Microscopic Traffic Flow Models Through Image Processing

Abstract Some decisions of urban traffic control are based on microscopic traffic flow simulation models. In each application parameters describing locally observed driving behaviour as well as parameters setting up the network have to be adjusted. To reduce the effort needed for calibration and validation this study deals with a parametrization method through image processing. Within the study one image processing system is applied to calibrate parameters of the microscopic traffic flow model of WIEDEMANN. A Simulated Annealing Algorithm is used to match observed and simulated values. Results of the calibration process are outlined using a partially signalized roundabout

Metastable states in a microscopic model of traffic flow

It is a well known fact that metastable states of very high throughput and hysteresis effects exist in traffic flow, which the simple cellular automaton model of traffic flow and its continuous generalization fail to reproduce. It is shown that the model can be generalized to give a one-parametric family of models, a part of which reproduces the metastable states and the hysteresis. The models that have that property and those that do not that are separated by a transition that can be clearly identified.

Macroscopic Modeling of Freeway Traffic Using an Artificial Neural Network

Traffic flow on freeways is a complex process that often is described by a set of highly nonlinear, dynamic equations in the form of a macroscopic traffic flow model. However, some of the existing macroscopic models have been found to exhibit instabilities in their behavior and often do not track real traffic data correctly. On the other hand, microscopic traffic flow models can yield more detailed and accurate representations of traffic flow but are computationally intensive and typically not suitable for real-time implementation. Nevertheless, such implementations are likely to be necessary for development and application of advanced traffic control concepts in intelligent vehicle-highway systems. The development of a multilayer feed-forward artificial neural network model to address the freeway traffic system identification problem is presented. The solution of this problem is viewed as an essential element of an effort to build an improved freeway traffic flow model for the purpose of developing real-time predictive control strategies for dynamic traffic systems. To study the initial feasibility of the proposed neural network approach for traffic system identification, a three-layer feed-forward neural network model has been developed to emulate an improved version of a well-known higher-order continuum traffic model. Simulation results show that the neural network model can capture the traffic dynamics of this model quite closely. Future research will attempt to attain similar levels of performance using real traffic data.

Traffic Control at Flagger-Operated Work Zones on Two-Lane Roads

To clarify the traffic flow characteristics and the signal timings at work zones controlled by flaggers, traffic flows at some work zones were measured. Using the measurement data a computer simulation program based on the microscopic traffic flow model was developed. The simulation results were in accordance with the measured results. Using the simulation program traffic behavior was analyzed for various work-zone conditions under the assumption that no heavy vehicles were mixed with the traffic. First, the delay characteristics at short work zones less than 200 m long were investigated. The summed delay of the acceleration delay and the speed-decrease delay was between 35 and 40 percent of the total delay. This suggests that those delays cannot be ignored in estimates of delay at short work zones. Next, three methods for estimating the green intervals under flaggers' control were presented: nomograph, regression equation, and analytical optimization. The first two methods were obtained directly from the simulation results, whereas the third was derived by describing both the queuing and the acceleration delays with a regression equation and minimizing the total delay. The analytically optimized green intervals diverged from the simulated ones as the traffic became heavier. The signal timings produced by a nomograph and a regression equation can be installed in pretimed traffic lights as a substitute for flaggers, and the optimized ones can be installed in actuated lights combined with traffic detectors.