No abstract

When traffic collisions occur on urban expressways, the consequences, including injuries, the loss of lives, and damage to properties, are more serious. However, the existing research on the severity of expressway traffic collisions has not been deeply explored. The purpose of this research was to investigate how various factors affect the severity of urban expressway collisions. The severity of urban expressway collisions was set as the dependent variable, which could be divided into three categories: slight collisions, severe collisions, and fatal collisions. Ten variables, including individual characteristics, collision characteristics, and road environment conditions, were selected as independent factors. Based on 975 valid urban expressway collisions, an ordered logistic regression model was established to evaluate the impacts of influence factors on the severity of these crashes. The results show that gender, collision modality, road pavement conditions, road surface conditions, and visibility are significant factors that affect the severity of urban expressway collisions. Females were more likely to be involved in more severe urban expressway collisions than males. For collisions involving pedestrians and non-motorized vehicles, the risk of more severe injury was 7.508 times higher than that associated with vehicle–vehicle collisions. The probability of more severe collisions on urban expressways with poor pavement conditions and wet surface conditions is greater than that on urban expressways with good pavement conditions and dry surface conditions. In addition, as visibility increases, the probability of more severe collisions on urban expressways gradually decreases. These results provide more effective strategies to reduce casualties as a result of urban expressway collisions.

No abstract

In the vehicular network, federated learning is an emerging paradigm to train deep learning models safely. However, the non‐identically independent distributed data collected by intelligent connected vehicles, expensive training overhead, and the existence of malicious participants significantly affect the efficiency of training and constrain the development of federated learning in the vehicular network. To address the above issues, this paper proposes a participant selection‐based asynchronous federated learning framework for an intelligent connected vehicle platoon. We proposed a platoon‐based asynchronous federated learning framework that effectively improves training efficiency through data share and asynchronous optimization. Also, We designed a signaling game‐based participant selection strategy that accurately identifies the potentially malicious participants by analyzing the Perfect Bayesian Equilibrium. The comparison experiment result shows the training model using our proposed scheme converges faster than the baseline scheme and gains approximately 10% higher accuracy.

Traffic flow prediction is an important function of intelligent transportation systems. Accurate prediction results facilitate traffic management to issue early congestion warnings so that drivers can avoid congested roads, thus directly reducing the average driving time of vehicles, which means less greenhouse gas emissions. However, traffic flow data has complex spatial and temporal correlations, which makes it challenging to predict traffic flow accurately. A Gated Recurrent Graph Convolutional Attention Network (GRGCAN) for traffic flow prediction is proposed to solve this problem. The model consists of three components with the same structure, each of which contains one temporal feature extractor and one spatial feature extractor. The temporal feature extractor first introduces a gated recurrent unit (GRU) and uses the hidden states of the GRU combined with an attention mechanism to adaptively assign weights to each time step. In the spatial feature extractor, a node attention mechanism is constructed to dynamically assigns weights to each sensor node, and it is fused with the graph convolution operation. In addition, a residual connection is introduced into the network to reduce the loss of features in the deep network. Experimental results of 1-h traffic flow prediction on two real-world datasets (PeMSD4 and PeMSD8) show that the mean absolute percentage error (MAPE) of the GRGCAN model is as low as 15.97% and 12.13%, and the prediction accuracy and computational efficiency are better than the baselines.