Search

Accurate traffic prediction is indispensable for relieving traffic congestion and people's daily trips. Nevertheless, accurate traffic flow prediction is still challenging due to the traffic network's complex and dynamic spatial and temporal dependencies. Most existing methods usually ignore the dynamicity of spatial dependencies or have limitations, as using the self-attention mechanism for capturing dynamic spatial dependencies is computation forbidden in large networks. In addition, there are both short- and long-range dynamic temporal dependencies, which are not well captured. To overcome these limitations, we propose an intelligent dynamic spatial and temporal graph convolutional system for traffic flow prediction. First, we propose a dynamic spatial block to capture the complex and dynamic spatial dependencies, which is computation-friendly. Next, we propose a dynamic temporal block to capture the complex and dynamic temporal dependencies, which well balances the short- and long-range dynamic temporal dependencies. We validate and analyze the performance of the proposed method through extensive experiments on two traffic datasets. Analysis of results demonstrates that our proposed model has better prediction performance than the state-of-art baselines. Compared with the best contrast methods, the proposed method improves by 2.28% and 8.01% in terms of the mean absolute error on PEMS04 and PEMS08 datasets. [ABSTRACT FROM AUTHOR]

Many highways are acquiring smart transportation systems to improve traffic efficiency, safety and management. Intelligent transportation systems can monitor traffic congestion by providing estimated time of arrival or suggesting a diversion route. However, due to the chaotic complexity of the highway traffic road network and the short-term mobility of the population, traffic flow prediction is affected by many complex factors, and the development of an effective traffic flow forecasting system is very challenging. This paper establishes a novel framework to enhance the prediction of highway traffic flow based on the integration of Complementary Ensemble Empirical Mode Decomposition and Gaussian Mixture Model (CEEMD-GMM). Firstly, the complementary ensemble empirical mode decomposition is used to decompose and reduce the noise of traffic flow data. The empirical modal components are combined by calculating the sample entropy (SE), which makes the time series stable and reduces the time cost of forecasting. Secondly, the histogram of oriented gradient (HoG) is computed to extract the distinct features of various traffic flows and enhance the traffic flow prediction. Furthermore, the highway traffic data is classified using Gaussian mixture model to realize the feature extraction of different types of highway traffics. The performance of the model is evaluated in terms of mean absolute error (MAE), root mean square error (RMSE), mean absolute percent error (MAPE), and average correlation. The model is extensively verified using traffic flow data under individual weather conditions and different traffic flow datasets. Compared with the traditional support vector machine (SVM), artificial neural network (ANN), random forest (RF) and Adaboost algorithms, the proposed model shows 2.90% MAE, 6.07% RMSE, 0.38% MAPE and average correlation coefficient of 0.74 for traffic flow detection. It is concluded that the model can predict the highways traffic flow with high accuracy as compared to other contemporary models. [ABSTRACT FROM AUTHOR]

Accurately estimated highway traffic flow info plays a decisive role in dynamic and real-time road management, planning, and preventing frequent/recurring traffic jams, traffic rule violations, and chain/fatal traffic accidents. Traffic flow information is extracted by processing raw camera images via vehicle detection and tracking algorithms. Object detectors including the Yolo, single-shot detector, and EfficientNet algorithms are used for vehicle detection; however, You only look once version 5 (Yolov5) has a clear advantage in terms of real-time performance. Due to this reason, the pre-trained Yolov5 models were utilized in the vehicle detection part, and in the vehicle tracking module, a novel tracker algorithm was developed using vehicle detection features. The performance of the proposed approach was measured by comparing it to the Kalman filter-based tracker. The evaluation results show that the proposed tracking approach outperformed the Kalman filter-based tracker with 5.82% (Buses), 2.24% (Cars), 36.50% (Trucks), and overall 2.58% better traffic counting accuracy for the 12 nighttime case study videos captured from the highways with different horizontal and vertical angle-of-views. [ABSTRACT FROM AUTHOR]

The rapid development of urbanization in China has contributed to traffic events, such as traffic accidents and delays. It is difficult to detect and resolve highway traffic congestion in a timely manner using traditional methods because they are slow, require a large number of workers, and require the installation of a large amount of monitoring equipment. Therefore, it is imperative to introduce advanced technology to address these challenges. Recently, deep learning technology has made significant breakthroughs and has been widely applied to various fields with satisfactory results. Deep learning is among the most important technologies for the detection and evaluation of traffic congestion, enabling the accurate detection of the state of the expressway network's traffic congestion, the evaluation of traffic congestion, and the prediction of possible traffic congestion. This enables management to formulate traffic dredging strategies in advance to prevent the negative impact of traffic congestion on normal traffic flow. This paper proposes a framework based on deep learning for next-generation highway traffic management. This framework selects traffic congestion indicators to construct an index model, and then constructs a deep learning model based on self-coding. It predicts and classifies highway traffic environment data and excavates sample data based on the characteristics of traffic parameters. As soon as traffic data were classified, a prediction model based on SoftMax was established to detect and predict highway traffic congestion. We conducted a traffic congestion analysis of the Shanghai expressway network based on the speed performance data obtained from the China Traffic Management Bureau. As a result of their research, we developed an index to measure highway traffic congestion. For traffic control and management organizations to function effectively, it is crucial to have an accurate and clear picture of traffic network operations. We evaluated the proposed framework using data gathered from highway monitoring scenes, and the results indicated that 98.6% of the data could be correctly detected. Using the prediction model based on SoftMax for expressway vehicles during peak hours, the accuracy was 92%, and the misjudgment rate was 8%. This study demonstrates that detecting and evaluating the state of highway traffic environments using deep learning has high accuracy, can be applied to actual highway traffic systems, and is extremely useful for detecting highway congestion. This framework is a promising solution for next-generation highway traffic management and provides accurate and timely traffic congestion detection and evaluation. [ABSTRACT FROM AUTHOR]

Accurate assessment of road conditions can effectively alleviate traffic congestion and guide people's travel plans, traffic control decisions of transportation departments, and formulation of traffic-related laws and regulations. This paper proposes a quantum particle swarm optimization (QPSO) and adaptive hybrid exponential smoothing with residual correction (AHES-RC) for the nonlinearity and randomness of traffic flow. In the proposed algorithm, the single–double-exponential smoothing method is mixed with adaptive weights to form AHES, a new method of mixing weights according to real-time traffic trend changes. After the residual correction of AHES is calculated by the extreme learning machine algorithm, the parameters of AHES-RC are optimized using QPSO to improve the prediction accuracy further. This paper comprehensively compares the QPSO-AHES-RC algorithm with other benchmark models through testing on 26 real-world data sets. The results show that the proposed algorithm can adaptive forecasting under various traffic conditions, yielding the best forecasting performance in terms of various forecast error metrics. Compared with advanced machine learning algorithms such as XGBoost and CatBoost, the mean RMSE and mean MAPE have been improved by over 20% on average. In addition, it is revealed that the real-time dynamic capture of traffic flow can effectively improve prediction accuracy. [ABSTRACT FROM AUTHOR]