Can Wang

This paper proposes event-triggered pinning graph reinforcement learning (EPGRL) for cooperative ramp merging in mixed traffic with connected and automated vehicles (CAVs) and human-driven vehicles. EPGRL represents vehicle interactions as a time-varying directed graph and uses a dual-stream graph encoder with temporal updating to capture evolving traffic couplings. A Lyapunov-based trigger activates cooperation when a stability-oriented surrogate index indicates loss of contractivity, while topology-aware pinning allocates control to influential CAVs. A PPO policy generates longitudinal and lateral actions, which are filtered by a safety shield. SUMO experiments across traffic demands and CAV penetration rates show that EPGRL improves travel time, fuel use, CO₂ emissions, and system-level time-to-collision relative to uncontrolled, infrastructure-based, and learning-based baselines. Event-timing and spatial analyses characterize the distribution of cooperative updates, while the control-update reduction rate ranges from 73.1% to 79.8% across representative settings. Ablation experiments further show that graph modeling, instability-aware triggering, and pinning selection make complementary contributions. These results indicate that selective, state-dependent cooperation can balance mobility, safety, energy use, and communication cost in simulated mixed-traffic merging.

Traffic sensors provide real-time measurements of current traffic conditions, whereas traffic management applications require forecasts of future traffic speed or flow. This study considers the incremental expansion of an operating sensor network, in which a pre-trained network-level forecasting model must predict traffic states at sensor locations that were absent during training. Many spatio-temporal graph neural networks rely on sensor-specific embeddings and graph connections learned from data-rich training networks. These representations are undefined for previously unseen locations, limiting the direct application of pre-trained adaptive-graph forecasters during sensor-network expansion. To address this cold-start problem, we propose support-conditioned sensor-adaptive meta-graph learning (SC-SAMG), which derives target-node representations and spatial dependencies from a short support period. The framework combines a support-set encoder, a task-specific graph learner, and first-order meta-learning to adapt the network-level forecaster using one to seven days of target observations. Experiments on four traffic benchmarks evaluate forecasts of speed or flow over the next 15–60 min under a leakage-controlled held-out-node protocol. SC-SAMG consistently outperforms fine-tuned gated recurrent unit (GRU), adaptive graph convolutional recurrent network (AGCRN), and diffusion convolutional recurrent neural network (DCRNN) baselines. It reduces mean absolute error (MAE) by up to 11% relative to the adaptive-graph baseline and by up to 7% relative to the diffusion convolutional baseline. These results demonstrate the potential of support-conditioned graph adaptation for incorporating previously unseen sensor locations into existing network-level traffic forecasting systems.

Efficient identification of key nodes is crucial to optimizing detector deployment and enhancing traffic monitoring in intelligent transportation systems. However, existing approaches often struggle to adapt to dynamic traffic variations, leading to suboptimal coverage and increased deployment costs. We propose bidirectional temporal convolutional graph attention networks (BTC-GATs) to address these limitations. This novel framework integrates bidirectional attention mechanisms to capture upstream and downstream dependencies, temporal convolutional networks for multiscale feature extraction, and graph attention networks for spatial information aggregation. BTC-GATs incorporates adaptive temporal modeling to capture nonlinear traffic dynamics, gradient-based variation analysis to quantify node influence, and a ranking mechanism that fuses attention coefficients with topological attributes to further enhance robustness and interoperability. In addition, a key node coverage study is conducted to examine the trade-off between accuracy and deployment efficiency. Extensive experiments on the California Highway PeMS04 dataset demonstrate that BTC-GATs outperforms benchmark methods in key node identification, offering superior accuracy and stability. Further analysis confirms its robustness under varying traffic conditions and initialization settings, highlighting its potential as a scalable, adaptive, and cost-effective solution for intelligent traffic monitoring. By facilitating efficient sensor placement, BTC-GATs contributes to improved data collection and congestion management in large-scale transportation networks.

In connected and automated environments, implementing feedback control on selected key connected and automated vehicles within a platoon can indirectly influence the operation of human-driven vehicles, ultimately optimizing overall traffic efficiency and safety. To enhance global traffic flow through centralized control of essential vehicles, an improved spectral clustering-based pinning control (ISC-PC) strategy is proposed in this paper. The strategy considers communication topology, vehicle individual features and vehicle interactive features to identify key control nodes, targeting them for pinning control to enhance traffic flow. A control algorithm is developed for these pinning vehicles, incorporating communication topology and vehicle dynamics, with a dual focus on safety and efficiency. Furthermore, the proposed ISC-PC strategy is validated through numerical simulations, employing both synthetic and real-world datasets, which include typical noises in traffic flow. Control effects are analyzed and compared under various pinning node identification methods and pinning rates. Experimental results indicate that the ISC-PC strategy is highly effective in reducing traffic oscillations and improving traffic efficiency and safety, underscoring its potential to maximize control effects under constrained control conditions.

High-quality traffic flow data is foundational to the study of traffic issues and practical engineering applications. The development of traffic flow detection methods has greatly facilitated the collection of traffic data; however, anomalies caused by factors such as equipment, networks, and environmental conditions present a common challenge. Traditional data restoration methods, which require extensive complete historical data as prior knowledge, are difficult to implement. In this study, we first analyze and categorize the causes of anomalous data, and introduce strategies for anomaly identification in multisource freeway data. Based on the unsupervised learning framework of generative adversarial network (GAN), and leveraging their advantages in data generation, we propose an auxiliary discrimination mechanism-based GAN (ADM-GAN) for the task of traffic flow data restoration. The optimization of the generator and discriminator and the auxiliary discrimination matrix, constructed from the generator’s output and the original data features, enhances the model’s utilization of original data. We tested our model using electronic toll collection traffic flow data from the G50 freeway near Suzhou, China. The results demonstrate that ADM-GAN outperforms other state-of-the-art methods in data restoration tasks across various data missing rates. This research offers an efficient and reliable method for freeway traffic data restoration.