Plug-in hybrid electric vehicles (PHEVs) with fuel and electricity have demonstrated the capability to reduce fuel consumption and emissions by adopting appropriate energy management strategies. In the existing energy management strategies, the dynamic programming (DP) -based energy management strategy (EMS) can realize the global optimization of the fuel consumption if the global vehicle-speed trajectory is known in advance. The global vehicle-speed trajectory can be obtained by applying GPS data of vehicles when the trip path is determined. However, for a trip path without GPS data, the global vehicle-speed trajectory is difficult to be gained. In this case, the DP-based EMS cannot be utilized to achieve the globally optimal fuel consumption, which is the issue discussed in this paper. This paper makes the following two contributions to solve this issue. First of all, the cell transmission model (CTM) of the road traffic flow and the vehicle kinematics are introduced to obtain the traffic speeds of road segments and the accelerations of the PHEV. On this basis, a hybrid trip model is presented to obtain the vehicle-speed trajectory for the trip path without GPS data. Next, a DP-based EMS with prediction horizon is proposed, and moreover, in order to improve its real-time implementation, a search range optimization algorithm of the state of charge (SOC) is designed to reduce the computational load of DP. In summary, we propose a computation-optimized DP-based EMS through applying the hybrid trip model. Finally, a simulation study is conducted for applying the proposed EMS to a practical trip path in Beijing road network. The results show that the hybrid trip model can effectively construct the vehicle-speed trajectory online and the average accuracy of the vehicle-speed trajectory is more than 78%. In addition, compared with the existing optimization algorithm for DP calculation, the SOC search range optimization algorithm can further reduce the calculation load of DP. More importantly, compared to the globally optimal DP-based EMS, although the proposed EMS make the fuel consumption grow less than 5.36%, it can be implemented in real-time. Moreover, compared with the existing real-time strategies, it can further reduce the fuel consumption and emissions. Thus, the proposed EMS can offer an effective solution for the PHEV applying it online in the trip path without GPS data.

For the on-line energy optimization problem of plug-in hybrid electric vehicles (P-HEVs), this paper proposes a heuristic dynamic programming-based on-line energy management strategy, to minimize the fuel consumption of the P-HEV. First of all, considering the uncertain nonlinear dynamic process of a vehicle in the actual traffic environment, we adopt the back propagation neural network (BPNN) to construct the dynamic model of the P-HEV. Then, on this basis, we utilize the heuristic dynamic programming to establish an energy management controller with the aim of minimizing energy consumption of the P-HEV. Moreover, the energy management controller is implemented by an on-line energy management strategy algorithm. To verify the effect of the controller, we employ a practical route in Beijing road network to simulate the BPNN model of the P-HEV and the proposed energy management strategy. The experimental results show several advantages of our strategy. Firstly, compared to the analytic model, the BPNN model can reflect the real dynamic process of the P-HEV with a higher precision. Secondly, the assigned torques by the strategy can effectively make the vehicle track the desired vehicle-speeds, and the tracking accuracy of the vehicle-speed is higher than 98%. Besides, on the premise of ensuring the real-time performance, the proposed strategy can further reduce the fuel consumption and emissions of the P-HEV when compared with the existing on-line energy management strategies, although its fuel consumption is more than that of the off-line global optimization energy management strategy by 4% approximately.

Risky and aggressive driving maneuvers are considered a significant indicator for traffic accident occurrence as well as they aggravate their severity. Traffic violations caused by such uncivilized driving behavior is a global issue. Studies in existing literature have used statistical analysis methods to explore key contributing factors toward aggressive driving and traffic violations. However, such methods are unable to capture latent correlations among predictor variables, and they also suffer from low prediction accuracies. This study aimed to comprehensively investigate different traffic violations using spatial analysis and machine learning methods in the city of Luzhou, China. Violations committed by taxi drivers are the focus of the current study since they constitute a significant proportion of total violations reported in the city. Georeferenced violation data for the year 2016 was obtained from the traffic police department. Detailed descriptive analysis is presented to summarize key statistics about various violation types. Results revealed that over-speeding was the most prevalent violation type observed in the study area. Frequency-based nearest neighborhood cluster methods in Arc map Geographic Information System (GIS) were used to develop hotspot maps for different violation types that are vital for prioritizing and conducting treatment alternatives efficiently. Finally, different machine learning (ML) methods, including decision tree, AdaBoost with a base estimator decision tree, and stack model, were employed to predict and classify each violation type. The proposed methods were compared based on different evaluation metrics like accuracy, F-1 measure, specificity, and log loss. Prediction results demonstrated the adequacy and robustness of proposed machine learning (ML) methods. However, a detailed comparative analysis showed that the stack model outperformed other models in terms of proposed evaluation metrics.

Short-term traffic state prediction has become an integral component of an advanced traveler information system (ATIS) in intelligent transportation systems (ITS). Accurate modeling and short-term traffic prediction are quite challenging due to its intricate characteristics, stochastic, and dynamic traffic processes. Existing works in this area follow different modeling approaches that are focused to fit speed, density, or the volume data. However, the accuracy of such modeling approaches has been frequently questioned, thereby traffic state prediction over the short-term from such methods inflicts an overfitting issue. We address this issue to accurately model short-term future traffic state prediction using state-of-the-art models via hyperparameter optimization. To do so, we focused on different machine learning classifiers such as local deep support vector machine (LD-SVM), decision jungles, multi-layers perceptron (MLP), and CN2 rule induction. Moreover, traffic states are evaluated using traffic attributes such as level of service (LOS) horizons and simple if–then rules at different time intervals. Our findings show that hyperparameter optimization via random sweep yielded superior results. The overall prediction performances obtained an average improvement by over 95%, such that the decision jungle and LD-SVM achieved an accuracy of 0.982 and 0.975, respectively. The experimental results show the robustness and superior performances of decision jungles (DJ) over other methods.

Short-term traffic speed prediction is vital for proactive traffic control, and is one of the integral components of an intelligent transportation system (ITS). Accurate prediction of short-term travel speed has numerous applications for traffic monitoring, route planning, as well as helping to relieve traffic congestion. Previous studies have attempted to approach this problem using statistical and conventional artificial intelligence (AI) methods without accounting for influence of data collection time-horizons. However, statistical methods have received widespread criticism concerning prediction accuracy performance, while traditional AI approaches have too shallow architecture to capture non-linear stochastics variations in traffic flow. Hence, this study aims to explore prediction of short-term traffic speed at multiple time-ahead intervals using data collected from loop detectors. A fast forest quantile regression (FFQR) via hyperparameters optimization was introduced for predicting short-term traffic speed prediction. FFQR is an ensemble machine learning model that combines several regression trees to improve speed prediction accuracy. The accuracy of short-term traffic speed prediction was compared using the FFQR model at different data collection time-horizons. Prediction results demonstrated the adequacy and robustness of the proposed approach under different scenarios. It was concluded that prediction performance of FFQR was significantly enhanced and robust, particularly at time intervals larger than 5 min. The findings also revealed that speed prediction error (in terms of quantiles loss) ranged between 0.58 and 1.18.

Non‐recurrent traffic incidents (accidents, stalled vehicles and spilled loads) often bring about traffic congestion and even secondary accidents. Detecting and positioning them quickly and accurately has important significance for early warning, timely incident‐disposal and speedy congestion‐evacuation. This study proposes a video‐based detecting and positioning method by analysing distribution characteristics of traffic states in a road segment. Each lane in the monitored segment is divided into a cluster of cells. Traffic parameters in each cell, including flow rate, average travel speed and average space occupancy, are obtained by detecting and tracking traffic objects (vehicles and spilled loads). On the basis of the parameters, traffic states in the cells are judged via a fuzzy‐identification method. For each congested cell, a feature vector is constructed by taking its state together with states of its upstream and downstream neighbouring cells in the same lane. Then, a support vector machine classifier is trained to detect incident point. If a cell is judged to be corresponding to an incident point at least for two successive time periods, an incident is detected and its position is calculated based on the identity number of the cell. Experiments prove the efficiency and practicability of the proposed method.