Frequent lane-changing maneuvers significantly increase the risk of traffic accidents on highways and constitute a critical factor affecting road safety. This study investigates the complex dynamic interactions that occur during vehicle lane changes in highway merging and diverging zones. Based on vehicle trajectory data captured from an autonomous aerial vehicle (AAV) perspective, a conflict prediction modeling approach is proposed that incorporates both trajectory features and spatiotemporal conflict characteristics. The method employs the attention mechanism of a graph attention network (GAT) to dynamically aggregate the spatial features of neighboring vehicles and integrates the self-attention mechanism of a Transformer to capture the temporal evolution of trajectories within a defined time window, thereby uncovering latent patterns and key distinguishing features of lane-changing conflicts in merging and diverging areas. Experimental results indicate that vehicles in merging zones exhibit higher lane-changing frequencies and greater lateral accelerations. Compared with baseline methods, the proposed GAT-Transformer-based conflict prediction model achieves superior performance, reaching a maximum prediction accuracy of 95.49% and an F1 score of 97.64%. These results underscore the model’s effectiveness in balancing predictive precision with comprehensive detection of true conflicts. Furthermore, the model consistently maintains prediction accuracies above 90% under varying penetration rates and demonstrates strong robustness and reliability across different levels of data availability.

In order to effectively solve the problem of electric taxi charging load prediction and reasonable charging behaviour discrimination, in this paper, we use taxi GPS trajectory data to mine the probability of operation behaviour in each area of the city, simulate the operation behaviour of a day by combining it with reinforcement learning ideas, obtain the optimal operation strategy through training, and count the spatial and temporal distributions and power values at the time of charging decision making, so as to predict the charging load of electric taxis. Experiments are carried out using taxi travel data in Shenzhen city centre. The results show that, in terms of taxi operation behaviour, the operation behaviour optimized by the DQN algorithm shows the optimal effect in terms of the passenger carrying time, mileage, and daily net income; in terms of the charging load distribution, the spatial charging demand of electric taxis in each area shows obvious differences, and the charging demand load located in the city centre area and close to the traffic hub is higher. In time, the peak charging demand is distributed around 3:00 to 4:00 and 14:00 to 15:00. Compared with the operating habits of drivers based on the Monte Carlo simulation, the DQN algorithm is able to optimise the efficiency and profitability of taxi drivers, which is more in line with the actual operating habits of drivers formed through accumulated experience, thus achieving a more accurate charging load distribution.

Against the backdrop of normalized multiple disturbances and uncertainties, the resilience of inland waterway shipping logistics corridors is of critical importance for maintaining regional logistics stability. To address the insufficient representation of corridor system attributes and the limited characterization of dynamic feedback among different dimensions, this study integrates corridor attributes with the process logic of the Driving Force–Pressure–State–Impact–Response (DPSIR) framework to develop a Channel Attributes–DPSIR framework. Dynamic weighting and structure-constrained dynamic Bayesian inference are further incorporated to characterize the time-varying features, coupling relationships, and uncertainties of resilience. Using the continuous corridor in the lower reaches of the Jinsha River as a case study, an empirical analysis is conducted based on data from 2015 to 2024. The results indicate that the overall corridor resilience improved steadily; driving and response resilience were relatively high, whereas state resilience showed slow improvement, acting as the primary bottleneck. The posterior results indicate that response has a stable feedback effect on drivers, impact-related losses can stimulate response investment, and pressure exerts a limited negative regulatory effect on state.