Hussein Dia

AN OBJECT-ORIENTED NEURAL NETWORK APPROACH TO SHORT-TERM TRAFFIC FORECASTING

Abstract This paper discusses an object-oriented neural network model that was developed for predicting short-term traffic conditions on a section of the Pacific Highway between Brisbane and the Gold Coast in Queensland, Australia. The feasibility of this approach is demonstrated through a time-lag recurrent network (TLRN) which was developed for predicting speed data up to 15 minutes into the future. The results obtained indicate that the TLRN is capable of predicting speed up to 5 minutes into the future with a high degree of accuracy (90–94%). Similar models, which were developed for predicting freeway travel times on the same facility, were successful in predicting travel times up to 15 minutes into the future with a similar degree of accuracy (93–95%). These results represent substantial improvements on conventional model performance and clearly demonstrate the feasibility of using the object-oriented approach for short-term traffic prediction.

Development and evaluation of neural network freeway incident detection models using field data

This paper discusses a multi-layer feedforward (MLF) neural network incident detection model that was developed and evaluated using field data. In contrast to published neural network incident detection models which relied on simulated or limited field data for model development and testing, the model described in this paper was trained and tested on a real-world data set of 100 incidents. The model uses speed, flow and occupancy data measured at dual stations, averaged across all lanes and only from time interval t. The off-line performance of the model is reported under both incident and non-incident conditions. The incident detection performance of the model is reported based on a validation-test data set of 40 incidents that were independent of the 60 incidents used for training. The false alarm rates of the model are evaluated based on non-incident data that were collected from a freeway section which was video-taped for a period of 33 days. A comparative evaluation between the neural network model and the incident detection model in operation on Melbournes freeways is also presented. The results of the comparative performance evaluation clearly demonstrate the substantial improvement in incident detection performance obtained by the neural network model. The paper also presents additional results that demonstrate how improvements in model performance can be achieved using variable decision thresholds. Finally, the models fault-tolerance under conditions of corrupt or missing data is investigated and the impact of loop detector failure/malfunction on the performance of the trained model is evaluated and discussed. The results presented in this paper provide a comprehensive evaluation of the developed model and confirm that neural network models can provide fast and reliable incident detection on freeways.

Neural Agent Car-Following Models

This paper presents a car-following model that was developed using a neural network approach for mapping perceptions to actions. The model has a similar formulation to the desired spacing models that do not consider reaction time or attempt to explain the behavioral aspects of car following. The models performance was evaluated based on field data and compared to a number of existing car-following models. The results showed that neural network models outperformed the Gipps and psychophysical family of car-following models. A qualitative drift behavior analysis also confirmed the findings. The model was validated at the microscopic and macroscopic levels, and the results showed very close agreement between field data and model outputs. Local and asymptotic stability analysis results also demonstrated the robustness of the model under mild and severe traffic disturbances

Development and evaluation of arterial incident detection models using fusion of simulated probe vehicle and loop detector data

This paper describes the development of neural network models for automatic incident detection on arterial roads, using simulated data derived from inductive loop detectors and probe vehicles. The work reported in this paper extends previous research by comparing the performance of various data fusion neural network architectures and assessing model performance for various probe vehicle penetration rates and loop detector configurations. Data from 108 incidents was collected from loop detectors and probe vehicles using a calibrated and validated traffic simulation model. The best performance was obtained for detector configurations found on most existing road networks, with a detection rate of 86%, false alarm rate of 0.36% and probe vehicle penetration rate of 20%. Fusion of speed data further improved performance, resulting in an incident detection rate of 90% and a false alarm rate of 0.5%. The results reported in this paper demonstrate the feasibility of developing advanced data fusion neural network architectures for detection of incidents on urban arterials using data from existing loop detector configurations and probe vehicles.

A reactive agent-based neural network car following model

This paper presents a car following model which was developed using reactive agent techniques based on a neural network approach for mapping perceptions to actions. The model has a similar formulation to the desired spacing models which do not consider reaction time or attempt to explain the behavioural aspects of car following. A number of error tests were used to compare the performance of the model against a number of established car following models. The results showed that simple back-propagation neural network models outperformed the Gipps and psychophysical family of car following models. A qualitative drift behaviour analysis also confirmed the findings. For microscopic validation, speed and position of individual vehicles computed from the model were compared to field data. Macroscopic validation involved comparison of the field data and model results for trajectories, average speed, density and volume. Model validation at the microscopic and macroscopic levels showed very close agreement between field data and model results.

Evaluation of discrete choice and neural network approaches for modelling driver compliance with traffic information

This article evaluates dynamic driver behaviour models that can be used, in the context of intelligent transport systems (ITS), to predict drivers’ compliance with traffic information. The inputs to this type of models comprise drivers’ individual socio-economic characteristics and other variables that may influence their compliance behaviour. The output is a binary integer representing whether drivers comply with travel advice or not. Two approaches are available for formulating this category of classification problems: discrete choice models and artificial neural networks (ANNs). The literature on this topic clearly points to the limitations of the discrete choice approach which suffers from assumptions of perfect information about travel conditions, infinite information processing capabilities of drivers and inability to model the uncertainty in driver decision making or the vagueness in information received from ITS devices. ANNs, on the other hand, are able to deal with complex non-linear relationships, are fault tolerant in producing acceptable results under imperfect inputs and are suitable for modelling reactive behaviour which is often described using rules, linking a perceived situation with appropriate action. This study aims to evaluate the performance of these two categories of models based on a common data set of driver behaviour, collected from a field behavioural survey on a congested commuting corridor in Brisbane, Australia. This article proposes the combination of fuzzy logic and neural networks as a viable approach for overcoming the limitations of existing algorithms by modelling drivers as heterogeneous individuals. The results showed superior performance for a neuro-fuzzy model over binary choice models in terms of classifying or predicting the categories of drivers most likely to comply (or not comply) with traffic advice. The accuracy of the proposed model, in terms of classification rate, ranged between 95 and 97% compared to 50–73% for the discrete choice models.

EVALUATION OF A DYNAMIC SIGNAL OPTIMISATION CONTROL MODEL USING TRAFFIC SIMULATION

The objective of this paper is to demonstrate the feasibility of implementing a traffic signal optimisation model to improve real-time operations of traffic control systems. Advanced computer algorithms and traffic optimisation techniques can provide benefits over existing systems by reducing delays, improving travel times and reducing environmental emissions. The feasibility of the proposed approach is demonstrated by interfacing the traffic signal optimisation model to a microscopic traffic simulation tool, which enabled the evaluation of the benefits of the algorithm using computers in a controlled environment without disrupting traffic conditions. The main advantage of the proposed algorithm is its ability to detect dynamic changes in traffic flow conditions by using short-term historical demand data obtained from upstream vehicle loop detectors. The experimental results for under-saturated traffic conditions showed that the algorithms performance was superior to optimal fixed time control. The results also confirmed that as traffic volumes reach saturated conditions, the performance of the algorithm decreased but remained better than what can be achieved by fixed time control systems.

Traffic Impact Assessment of Incident Management Strategies

This paper presents results from a simulation study which aimed to quantify the impacts of incident management strategies. The evaluation was based on a large-scale micro-simulation model covering an area approximately 122 square kilometres, including 43 kilometres of motorway and about 85 kilometres of surface roads on the Gold Coast, Australia. The study examined the effectiveness of selected incident management strategies including ramp metering, VMS information dissemination combined with route diversions, and variable speed limit systems. The provision of VMS information on the motorway and dynamic adjustment of signal timings on the diversion route resulted in equilibrium conditions and balanced distribution of traffic on both the normal and diversion routes when the optimal diversion rate was 30 percent. This resulted in reduction of delays by 8.8 percent, decrease in number of stops by 22 percent, and decrease in travel times by 3.3 percent. An important finding was that these benefits are only realised when the two incident management responses (VMS route diversion and dynamic traffic signal plans on surface roads) are implemented at the same time. Combined, their impact was such that they resulted in restoration of traffic conditions to the pre-incident situation. The paper also reports on a preliminary investigation into variable speed limits (VSL) as a means to reduce the negative impacts of incidents. The results showed that VSL had the potential to provide an 11 percent improvement in efficiency and also contributed to improving safety by homogenising the flow in higher speed regimes.

Comparative evaluation of power-based environmental emissions models

This paper presents findings from a comparative evaluation of two Australian power-based models of fuel consumption and environmental emissions using traffic simulation. The simulation approach provides a number of advantages for this type of work through detailed modelling of road geometry and grade, driver behaviour, vehicle performance and traffic parameters. The power-based models were interfaced to a microscopic traffic simulator and applied to each vehicle on the road network. Second-by-second fuel consumption and pollutant emissions were estimated on a section-by-section and network-wide basis. The simulation study examined a number of abstract vehicle types and gathered detailed information about their speed profiles, acceleration, deceleration, fuel consumption and pollutant emissions. The results showed that the fuel consumption outputs were similar but the emission estimates were substantially different for the two models. The paper also compared the outputs of the two models against various standard tests. The results showed that both models provided higher rates of pollutant emissions when compared with available standard test data. The paper concludes with some recommendations for further research to enhance the performance of the models

Impact of Driving Behaviour on Emissions and Road Network Performance

This paper presents findings from a simulation-based comparative evaluation of driving behaviours and their impacts on road safety, environmental quality and network efficiency. Driving behaviour was represented by driver speed, acceleration, lane changing and gap acceptance actions. A fourmode elemental emissions model was used to collect second-bysecond data on fuel consumption and CO2 emissions. Surrogate measures of safety, expressed in terms of the number of lane changes and severe decelerations, were used to describe the degree of safety in the simulation experiments. Aggressive drivers were found to be 35 times more likely to be involved in a crash on the motorway, and two times more likely to be involved in a crash on the urban network. The results for the motorway simulations also showed that aggressive drivers achieved only a 3.8 percent reduction in travel times (62 seconds on a 26 minute trip) at the expense of 85 percent more lane changes and 332 percent increase in fuel consumption and CO2 emissions. The reduction in travel times for urban conditions was lower at around 1.6 percent (7 seconds on a 434 second trip) at the expense of 300 percent more lane changes and 138 percent increase in fuel consumption and CO2 emissions. Sensitivity analysis of the impacts of varying proportions of drivers was also conducted. The results showed that the negative impacts of aggressive driving behaviour outweigh by a factor of three any benefits that can be obtained through reductions in travel times.