This study proposed an adaptive information release strategy of VMS (variable message signs) in road network based on the real-time response of the passenger flow, in order to improve the travel efficiency and reduce the safety risk of pedestrians in the competition area under low temperature environment. First of all, from the perspective of information intervention, the dynamic feedback mechanism between the VMS information release and the crowded state of the passenger flow was established, and the dynamic optimization model of the VMS information release layout was formulated. The in response to the strategy, a scene simulation method based on multi-agents was proposed. A cold competition area was taken as an example to simulate the ingress and egress scenes with and without the VMS information release strategy. The results show that the proportion of short-term passenger flow increases, compared with that without the VMS information release strategy. For the ingress scene, the ave-rage walking time of pedestrians can be reduced by about 2.6%, and the maximum road congestion can be reduced by about 20.45%; while for the egress scene, the average walking time of pedestrians can be reduced by about 7.0%, and the maximum road congestion can be reduced by about 10.51%. The research can provide theoretical foundation and data support for the competition area managers to control passenger flow and improve the experience of spectating experience.

In order to explore the activity pattern and regularity of public transport passengers, this study constructed multi-day passenger travel activity sequences using three weeks smart card data in Beijing in October 2020. The frequent activity pattern sequences of passengers were mined through the PrefixSpan algorithm, and the similarity measure method of activity patterns was defined based on the longest common subsequence. The day-to-day activity sequence similarity of individual and activity pattern similarities among different passengers were calculated respectively, and passengers were classified according to activity pattern similarities among passengers by using the hierarchical clustering algorithm. The results show that the similarity between workdays and weekends is significantly lower than that within workdays or weekends. In workdays, the activity sequence similarity between Friday and the other days is low. Meanwhile, the activity sequence similarity of the same days in different weeks is high. The result of hierarchical clustering shows that there are four typical activity patterns, including entertainment and shopping orientation, life orientation, work orientation and personal affair orientation. Moreover, the day-to-day activity sequence similarity of passenger with work orientation pattern is higher than that of passenger with other activity patterns. The research results in this paper are helpful to scientifically formulate accurate public transport operation management and service policies.

In response to the concept of land-use integration and creating micro-center around metro station in Beijing, this study extracted 23 quantitative indicators from public passenger flow, road network design, population density and land diversity to quantitatively analyze the built environment and travel characteristics of the non-motorized influence area based on multi-source big data. The connection characteristics of shared bicycles were taken into particular consideration. In order to compensate for the shortcomings of determining the influence range of metro stations by the traveler’s walking time, a classification model incorporating principal component analysis and K-means clustering was proposed to define the non-motorized influence area. Taking Beijing as an example, the study divided the metro stations into 4 clusters: inefficient connection-weak connectivity-residence oriented, efficient connection-high connectivity-balanced, efficient connection-weak connectivity-mixed, and efficient connection-high connectivity-work oriented. In order to verify the rationality of the clustering, the spatial auto-correlation was used to judge the spatial dependence of indicators. The results show that the spatial distributions of clusters 1, 3 and 4 do not differ significantly from the random model, while cluster 2 efficient connection-high connectivity-balanced stations has auto-correlation characteristics in space. Finally, based on the clustering results, the non-motorized influence areas of the metro stations were delineated as 2000, 1600, 1600, and 1700 m, respectively. The clarification of the non-motorized influence range of different metro station types can help urban planners determine the scope of micro-center construction and also lay the foundation for transportoriented development of urban in the future.

As a green travel option, the bicycle sharing system has seen rapid development in recent years. Where to put the repaired bicycles and new bicycles becomes a real problem. It is necessary to help the bicycle sharing system rebalance the supply and demand. Moreover, it is also important to ensure the quickly use for new bicycles. Based on the realistic demand for bicycle sharing, this paper designs a strategy to select the location of bicycle sharing delivery points. The strategy considers the current supply and demand as well as travel intensity. Based on mobile signaling data and bicycle sharing data, the potential demand for bicycle sharing use is identified firstly. Next, the distribution of bicycle sharing is recognized. Then the supply and demand of bicycle sharing are analyzed. Finally, travel hotspots are identified and the location of bicycle sharing delivery points is extracted. Taking Beijing as a case study, more effective delivery points within the fifth ring road were identified. The application of the method can help improve the operational efficiency of the bicycle sharing system and promote the sustainable development of urban transportation.

Driving style is the external expression of driving behavior. Drivers with aggressive style tend to engage in more frequent risky driving operations, intensifying interactions between vehicles and affecting lane-changing safety. Identifying a driver’s driving style before executing a lane-changing can effectively constrain driver’s behavior through personalized warning information. This paper proposed the SHAP-XGBoost method, which considers lane-changing game in a connected environment, aiming to achieve the real-time recognition of driving styles during the lane-changing intention phase. Firstly, the fluctuation degree of individual behavior and gaming behavior during the lane-changing intention was used as input feature variables, and the driving style was marked by correlation analysis, principal component analysis, and four different clustering methods. Next, the proposed SHAP-XGBoost model was used to select key features for training the driving style recognition model, and online recognition was completed through a sliding window. Finally, experiments were conducted using the HighD dataset. Results show that: compared with clustering methods based on centroid distance, connectivity and density distribution, spectral clustering based on graph theory principles can better label driving styles based on the morphology of the input feature variables; using the proposed SHAP-XGBoost model with 14 key features for driving style recognition can improve online recognition efficiency without loss of accuracy, and the driving style recognition accuracy is up to 99%; simultaneously incorporating individual features and gaming features as inputs to the model can improve the accuracy of driving style labeling and recognition. The research results can be used to support personalized lane-changing decisions and early warnings.

Effective traffic management and congestion reduction heavily rely on accurate traffic flow prediction. Existing prediction methods, such as Markov, ARIMA, STANN, GLSTM, and DCRNN models, often face challenges because they rely on fixed spatial relationships, leading to limited long-term prediction accuracy. To address these shortcomings, this study proposes the Impedance-Spatio-Temporal Topological Network (Impedance-STTN) prediction model. The Impedance-STTN model integrates K-medoids clustering for data analysis, generating a real-time impedance matrix from impedance functions, traffic big data, and real-time flow data. This approach captures dynamic node relationships within the spatio-temporal network, enhancing prediction accuracy. Experimental results demonstrate the superior predictive performance of the Impedance-STTN model, achieving accuracies of 94.79%, 93.78%, and 93.11% in 5 min, 15 min, and 30 min predictions, respectively. These results outperform existing models, especially in long-term predictions. The findings underscore the model's high accuracy and effectiveness across varying prediction durations, marking a significant advancement in traffic flow prediction. This suggests promising avenues for future research and practical applications.

The efficient, reliable, and sustainable nature of a transportation system is a prerequisite to support the development of urban agglomeration. This paper proposes network modeling and resilience assessment methods for public transportation in urban agglomerations. A multi-layer network is constructed. With the identification of the key nodes in a multi-modal transportation network (MMTN), a resilience assessment method is proposed that considers two phases: absorption and recovery after an attack. The Beijing-Tianjin-Hebei urban agglomeration network is taken as a case study. The results show that the attack on key nodes brings more influence to MMTN than random attacks. More attention is suggested to be paid to the larger hub-type stations in operation and management. The proposed method can be applied in different types of urban agglomerations and serve as technical support for reducing the disorder and imbalance of MMTN.