IET Intelligent Transport Systems最新文献

On the basis of a generative adversarial network (GAN) and a convolutional neural network (CNN), this work proposes an ER-GAN-CNN to forecast the traffic flow in the presence of missing data by improving GAN. Due to the occurrence of emergencies and the fault of the relevant equipment, the equipment for passenger flow detection may lose some data, which would have negative impacts on passenger flow prediction. In order to cope with such situations, GAN is introduced to make up for the missing data in this paper. On the basis of complete data and CNN, inception blocks are built thereafter to further predict the passenger flow/traffic flow. The accuracy of the prediction of passenger flow is significantly improved with the help of the ER-GAN-CNN, which is able to provide more accurate and rapid traffic guidance for the drivers.

The current moving block systems still depend on wayside interlocking systems and zone controllers, resulting in complex control flow and limited control efficiency. While recent train-centric solutions have simplified the system and enhanced line capacity, they still require explicit resource requests from the wayside infrastructure controller and necessitate storing all routes in onboard equipment. These limitations constrain system performance and maintainability. This study introduces an autonomous train control principle for fully train-centric route generation and track resource management, eliminating reliance on wayside controllers. The proposed system models track layouts as directed graphs and generates routes through route factorisation and composition, ensuring compliance with railway safety and operational requirements. By utilising train-to-train coordinate transformations, ATCS enables direct management of track resources between trains without intermediaries, significantly improving the system's performance. Furthermore, a novel braking model is introduced, optimising headway distances and improving track utilisation. The proposed principle is evaluated on an actual railway track layout in Korea, and the results demonstrate its feasibility, achieving shorter headways, improved track capacity, and enhanced system maintainability and flexibility when compared to conventional CBTC and train-centric CBTC systems.

Traffic forecasting plays a pivotal role in the advancement of intelligent transportation systems, with significant implications for congestion alleviation and optimal route planning. Existing approaches typically focus on capturing the temporal dynamics of traffic states and the spatial dependencies across road networks to improve prediction accuracy. Nevertheless, two noteworthy limitations persist in these approaches: (1) A lack of consideration for the interaction between spatiotemporal features over varying time scales, which impedes the effective utilization of traffic state information for forecasting future conditions. (2) The inherent stochasticity and distributional imbalances in traffic flow, which introduce uncertainty and contribute to overfitting issues in deep learning models. To address these challenges, we propose a novel method, the heterogeneous-scale multi-graph convolution networks based on kernel density estimation (KDE-HSMGCN). This method integrates two core components: the frequency feature layer and the heterogeneous-scale spatiotemporal layers. The frequency feature layer employs a mapping network to learn and equalize traffic flow distributions, mitigating the effects of distribution imbalance and overfitting during model training. The heterogeneous-scale spatiotemporal layers utilize stacked spatiotemporal layers to capture traffic state information across varying time scales. Experimental evaluations on two diverse traffic datasets demonstrate the superior performance of KDE-HSMGCN in medium and long-term forecasting scenarios.

Urban traffic speed prediction with high precision is the unremitting pursuit of intelligent transportation systems. The fundamental challenges of traffic speed prediction lie in the accurate modelling of the complex temporal and spatial correlations of transportation systems. Among all the methods, the hybrid “GNN + RNN” models have achieved state-of-the-art results. However, these methods still cannot address the following two challenges. First, in addition to the topology of road networks, the traffic speed could be affected by a variety of other factors, such as road functionality and weather. Second, in addition to predicting traffic speed, it is necessary to diagnose the causes of the prediction results. In this paper, we propose a multi-graph attentive network (MGAN), to predict and diagnose urban traffic speed. We create GNN model by using multiple graphs to encode the factors affecting them from various aspects. And we design a hierarchical attention mechanism to organize and pinpoint the fine-grained effects of different affecting factors for diagnosing the prediction results. The experimental results demonstrate that MGAN achieves state-of-the-art prediction performance on two real-world datasets, outperforming the strongest baseline by at least $��5.94�\%�$ across three prediction horizons, and is able to intuitively diagnose the prediction results.

Accurate platform-based passenger flow prediction based on deep learning technology becomes crucial for efficient operation and management; in particular, the prediction integrating external weather factors, temporal dependencies and spatial features is desired but has not been addressed. This paper is different from the previous station-based passenger flow prediction, but reconstructs data recorded by the Automatic Fare Collection System (AFCS) for platform-based prediction. Existing deep learning techniques often struggle with issues such as high computational cost in traffic flow prediction. To address these issues, a novel passenger flow prediction model is proposed that integrates convolutional neural networks (CNN) with bi-directional long short-term memory networks (BILSTM) and an attention mechanism (CNN-BILSTM-Attention). The proposed model takes preprocessed numerical weather features, temporal and spatial features as input. The CNN extracts spatial patterns from passenger flow data, the BILSTM captures temporal dependencies and the attention mechanism dynamically adjusts the importance weights of features at different time slots. By integrating these components, the model effectively captures spatiotemporal patterns while accounting for weather impacts. Experimental results demonstrate that the proposed approach outputs an efficient and accurate prediction.

The energy consumption of the subway has attracted much attention. Applying bidirectional converter substations (BCS) and researching energy-efficient train control (EETC) strategies can effectively reduce the energy consumption of the subway system. This paper analyzes the coupling model of power supply-train operation with rectifier substations (RS) and BCS. To minimize the energy consumption of substations, an optimal control problem model of EETC is established, and a multi-stage dynamic programming algorithm with state space reduction is designed to solve the train energy-saving speed profile. The EETC results of different line conditions with RS and BCS are presented. The results indicate that the EETC changes with the type of substations, where trains with BCS adopt regenerative braking conditions matched with the inverter turn-on voltage to feed back energy. The relationship between train running time and energy consumption is analyzed, showing that EETC with BCS has superior energy-saving effects and operational efficiency to EETC with RS. Results demonstrate the effectiveness and energy-saving effects of the optimization methods presented in this paper.

This paper introduces a dynamic optimisation system that enhances the management of train delays within automatic train operation (ATO) systems, utilising an innovative integration of shrinking-horizon model predictive control (SH-MPC) with genetic algorithms (GA). This research focuses on optimising train trajectories to efficiently handle various delay scenarios, from temporary speed restrictions to significant halts, ensuring both energy efficiency and punctuality. The proposed SH-MPC addresses diverse delay situations in real time, while the integration with GA overcomes the limitations of long horizon forecasting. The simulation of multiple trains on a real route demonstrates the robustness of the proposed system in adhering to scheduled timetables while reducing energy consumption.

This paper investigates the integration of rate-splitting multiple access (RSMA) into cellular vehicle-to-everything (C-V2X) networks to enhance resource allocation and interference management in decentralized, ad-hoc vehicular communication environments. C-V2X facilitates communication among vehicles, infrastructure, and pedestrians, and traditionally relies on orthogonal frequency division multiple access (OFDMA). However, OFDMA's rigidity limits its effectiveness under dynamic interference and imperfect channel state information (CSI) conditions typical of vehicular networks. RSMA, which blends features of spatial division multiple access (SDMA) and non-orthogonal multiple access (NOMA), provides a more adaptive framework by splitting messages into common and private parts, thereby improving spectral efficiency and interference handling. To assess RSMA's applicability, the LTEV2Vsim simulator was extended to include RSMA functionality, incorporating features such as reputation-based grouping, group-wise resource synchronization, and simplified beamforming. A dynamic grouping algorithm selects high-reputation vehicles as transmission leaders to form multi-vehicle groups of varying sizes for RSMA-based transmission. For interference modeling, self-interference is excluded from SINR calculations, and beamforming-based inter-vehicle interference is approximated. Simulation results reveal that RSMA outperforms OFDMA in terms of spectral efficiency and adaptability, particularly under conditions of incomplete CSI and varying interference. The findings confirm RSMA's suitability for complex and fast-changing vehicular environments, indicating its potential as a robust multiple access scheme for future C-V2X deployments.

The prediction of dynamic origin-destination (OD) data is critical for facilitating real-time traffic management across traffic networks. Despite numerous efforts to integrate the temporal and spatial characteristics of OD data to capture the nonlinearity and high dynamics of traffic flow, prior studies usually rely on link-level or region-level data for model construction. The temporal relationships among origin traffic flow, destination traffic flow, and OD flow remain insufficiently understood. To address this gap, we propose a novel definition of dynamic OD data using real-world OD datasets. Our framework can incorporate different temporal distributions for each OD pair. Additionally, the framework ensures flow conservation from either the origin or the destination perspective. The performance of the proposed framework is validated through numerical studies using real-world electronic toll collection (ETC) gantry data. A multi-task long short-term memory (LSTM) model predicts OD flows, and both the predictions and the resulting destination traffic distributions are statistically indistinguishable from the observed values. Furthermore, this approach enables the prediction of arrival volumes before trip completion, offering valuable insights for real-time traffic management.

Accurate traffic speed prediction is important in modern society for its effectiveness in route navigation, estimated time of arrival calculations and other practical applications. As the road network is complicated, traffic speed exhibits high-order correlations among regions, namely many-to-many spatial correlations, while also displaying long-term temporal dependencies. However, existing studies have not effectively modelled these characteristics. In this context, this study proposes a self-attention enhanced hypergraph convolution network (SE-HCN) for accurate speed prediction. The proposed SE-HCN consists of four modules. Specifically, we design a relation extraction module, which can obtain the similarity of road sections from geographical information and clustering. Subsequently, the model contains a spatial correlation hypergraph convolutional module and a long-term temporal dependencies transformer module to capture spatio-temporal features comprehensively. Two public real-world datasets - PeMSBAY and PeMSD7-M - were tested to validate the model's performance, and the result demonstrates that our approach achieved state-of-the-art performance.