[Objectives] The scale, distribution, travel mode structure, and traffic flow of passenger travel demand are the results of spatial interactions within the human social economy across different locations. The complexity of the social and economic operation systems dictates that travel demand prediction must start from the urban system to address the technical challenges of current travel demand forecasting. This paper analyzes the systematic nature of urban transportation and proposes an integrated simulation technology framework that incorporates land, population, housing, and transportation. It also summarizes traffic demand simulation and prediction technology based on urban systems and develops China's first urban system travel demand forecasting technology platform. [Methods] This technology covers sub-modules such as transportation demand distribution, transportation mode share and path allocation, land use simulation, population and employment distribution, real estate price, and carbon emissions to reflect the complete urban system. It includes a series of sub-module variables, including generalized travel cost, location accessibility, real estate price, job-housing relationship coefficients, and land use mixing degrees, to reflect the interactions among subsystems and the time lag effect. Additionally, core algorithms of sub-modules are designed to achieve urban system simulation and prediction. Using Beijing as a case study, the application of this technology platform is demonstrated. A comparison between the actual and simulated values for 2020 shows that the accuracy of simulated results for travel demand, traffic congestion situation, land use, and population distribution is above 85%. [Results] Applying this platform to Beijing, the travel demand, traffic flow, congestion index, population distribution, and land use projections for 2030 were predicted. According to the forecast results, from 2020 to 2030, the total number of traffic trips in Beijing will show a generally stable and slowly declining trend, with strong centripetal characteristics spatially, and trips within each suburb will become more balanced. There will be a slight decrease in the proportion of public transportation travel, a slight reduction in residents' average travel time, and more severe congestion compared to 2020. The expansion of land for residential areas, roads and transportation facilities, green spaces and squares, and commercial services will be more obvious. Resident population will show steady fluctuations, with finger-like extensions along major transportation corridors. [Conclusion] Overall, this paper advances urban transportation theory, innovates urban transportation simulation forecasting methods, and provides new technical support for urban and rural planning and urban transportation planning.

[Significance] Cities globally face increasingly frequent multi-hazard risks, driving them pursuing more sustainable and resilient urban transportation systems. This paper presents a comprehensive systematic literature review of the application of spatial-temporal data in transportation system resilience studies. It highlights the pivotal role of spatial-temporal big data in understanding and enhancing the resilience of urban transportation systems under various hazard scenarios. Spatial-temporal big data, characterized by high temporal resolution and fine spatial granularity, has been increasingly applied to the field of transportation system resilience, providing essential support for decision-makers. [Progress] This study reveals two significant findings: Firstly, quantitative analysis of transportation system resilience is one of the most widely applied uses of spatial-temporal big data. However, real-time monitoring and early warning explorations are relatively rare. Most studies remain at the modelling and numerical simulation stage, indicating a need for more empirical studies using multi-source spatial-temporal big data. Moreover, compared to English literature, Chinese transportation system resilience studies are primarily qualitative and lack empirical research, indicating divergent research emphases between domestic and international scholars. Secondly, high-quality, multi-source spatial-temporal big data could facilitate more comprehensive spatial analysis in transportation system resilience studies. Improved data quality allows for deeper exploration from a microscopic perspective, focusing on individual behaviors and aligning closely with real-world needs. The concept of resilience has evolved from its previous post-disaster focus to a comprehensive life-cycle perspective encompassing pre-, during-, and post-disaster phases, transforming the study framework for transportation system resilience. [Prospect] As spatial-temporal big data technology advances and new transportation modes emerge, more innovations and breakthroughs in transportation system resilience studies are expected. Future research should further explore and utilize the potential of spatial-temporal big data in this field, amplifying the policy ramifications of abrupt-onset occurrences. Increased emphasis should be placed on research conducted at the scale of urban agglomerations. Simultaneously, a nuanced examination from a microscopic perspective is imperative to dissect the underlying causes and mechanisms contributing to variations in resilience among distinct groups. Despite the significant progress in transportation system resilience studies, there are still challenges in data collection, processing, and analysis. As technology progresses, researchers should leverage advanced algorithms, platforms, and tools to enhance data processing capabilities and analytical precision, facilitating more complex and detailed studies on transportation system resilience. This will provide a scientific basis for planning and managing urban transportation systems, significantly contributing to the overall resilience and sustainable development of cities.

[Objectives] Understanding the spatial and temporal characteristics of different types of taxis operations is crucial for transport planning and management. In this paper, a novel research model, Taxi Mobility Chain (TMC), is constructed by stringing together consecutive movements of individual taxis using order data from traditional taxis and e-hailing cars in Xiamen City. The introduction of TMC addresses the limitation of traditional research that uses isolated OD flow as the analysis unit to some extent, and provides richer background information, which can connect with more intelligent methods and has better universality. [Methods] Based on the obtained TMCs, we designed corresponding indicators as analytical tools from the dimensions of taxi usage and TMC characteristics. These tools are used for the macroscopic description of the usage of traditional taxis and e-hailing cars and the quantitative computation of the spatial morphological characteristics of TMCs. In particular, with reference to the word embedding technology in the field of Natural Language Processing (NLP), we explore the spatiotemporal pattern characteristics formed during taxi operations by using clustering algorithms after embedding the cellular areas of the study area grid. [Results] By comparing and analyzing two kinds of taxis with different operation modes ("vehicles looking for people" and "people looking for vehicles"), the results of the indicator analysis module mainly indicate that the average daily number of orders received by a single taxi does not differ much between weekdays and non-weekdays, and the individual utilization rate is relatively stable. However, the drivers of traditional taxis have more autonomy in taking orders, more consecutive orders, and higher work intensity. The results of the cluster analysis module mainly indicate that traditional taxi operations show stronger aggregation and are more inclined to shuttle back and forth in some hotspots. E-hailing cars are based on individual travelling needs, with boarding and alighting points completely decided by passengers, resulting in a larger spatial service scope. In addition, when the number of clusters is equal to 2, the clustering results do not differ much between weekdays and non-weekdays, and roughly correspond to the primary and secondary operation areas of the two types of taxis. When the number of clusters is equal to 6, traditional taxis with higher autonomy are clustered in high outbound travel demand areas on non-weekdays. The overall clustering results of traditional taxis are more correlated with the spatial distribution of the high-density road network, while the clustering results of e-hailing cars are more correlated with administrative districts. As the number of clusters increases from 2 to 18, the clustering characteristics of traditional taxis operations become more prominent. [Conclusions] Using TMC-based cluster analysis tools, the thesis can better identify the spatiotemporal characteristics of taxi operations and reveal the status and role of different types of taxis in urban transport.

[Objectives] Urban public transportation service quality is an important factor affecting residents' travel choices and quality of life, but the current development and reform of urban public transportation in China still has shortcomings, and it is necessary to incorporate public perception into the decision-making basis and improve service quality from the perspective of residents. Previous studies have two main limitations: first, they rely on traditional analysis methods based on traffic surveys, which fail to capture the regional differences in perceived service quality; second, they use big data from social media platforms, which are prone to information bias, polarization, and other issues, and do not reflect the public's real needs. Moreover, they mostly focus on public opinion analysis, without providing specific and feasible optimization paths. [Methods] To address these gaps, this paper proposes a method that combines public network participation and semantic analysis. It uses internet big data to extract online messages related to urban public transportation from the online interactive platform between government and citizens and analyzes their spatiotemporal features and perceived service quality. It also conducts spatial analysis and explores the service efficiency of the public transportation system in relation to the transportation facility distribution. Based on this, it offers optimization suggestions. The paper selects Wuhan as a case study, which is one of the national central cities and an important megacity in the middle reaches of the Yangtze River. The urban development area in Wuhan is a key zone for urbanization and a major hub for public travel activities, covering 15 functional zones. It has a complete public transportation facility allocation, including all the subway lines and stations, and most of the bus lines and stations in the city. [Results] The main findings are as follows: (1) The quality of public network participation data can reflect the spatiotemporal patterns of actual travel activities and has high credibility; (2) The emotional expression of the public varies across individuals and regions and the perceived service quality dimensions can be categorized into five topics: "public transportation planning and construction", "public transportation travel conditions", "residential community bus configuration", "public transportation route setting", and "public transportation operation service". Furthermore, the perceived service quality exhibits spatial imbalance and agglomeration; (3) Corresponding optimization suggestions are made for the road system in the main urban area, subway stations in the far urban area, and bus routes at the junction of the main urban area and far urban area. [Conclusions] The research results of this paper provide a new method for fine-grained identification and optimization of spatial differences in urban public transportation perceived service quality, and also demonstrate the application value of public network participation data in facilitating government decision-making.

[Objectives] Identifying regular travel groups on freeways is a prerequisite for accurately guiding travel demand and providing personalized travel services. [Methods] Existing research mainly focuses on urban transportation, extracting travel characteristics based on time-frequency indicators to identify regular travel groups. However, the identification of regular users for freeway systems, characterized by large travel time spans and varying travel times, has not been fully explored. [Results] This study, based on toll data from the Guangxi Zhuang Autonomous Region freeway network, analyzes user trajectories between toll booth entry and exit points as signal inputs. Using Fourier transform, the variance of the top three amplitudes corresponding to periodic signals is calculated as a special index to estimate whether trips are regular. A Fourier transform-based 'trajectory plotting-spectral transform-period analysis' method is proposed for identifying and analyzing cyclical freeway users, with the Guangxi Zhuang Autonomous Region as the case study. [Conclusions] The results show that: (1) Freeway travel records in Guangxi Zhuang Autonomous Region exibit a head effect, with 20% of vehicles accounting for 55%~60% of toll records. (2) At a periodic variance threshold of 0.5, about 12% of passenger and cargo vehicles are identified as regular users; at a threshold of 1.0, this increases to about 18%; and at a threshold of 2.0, it rises to approximately 25%. (3) Passenger cars with frequent travel patterns do not necessarily exhibit periodicity, while vehicles (both passenger and cargo) with periodic patterns typically have shorter travel distances and more stable travel times. The findings of this study provide a reference for expanding methods of identifying regular freeway users and exploring the periodic characteristics of roadway user trips.

[Objectives] The ultra-wideband ranging errors in underground narrow spaces show a significant heavy-tailed distribution. The Gaussian mixture model is more closer to the empirical distribution than a simple Gaussian probability envelope. The Protection Level (PL) obtained through traditional bounding distribution models are overconservative, which reduces system usability. [Methods] To enhance the system usability, an overbounding framework based on Gaussian Mixture Model (GMM) is employed to handle the Time of Arrival (TOA)-based distance measurements obtained from the ultra-wideband ranging system. Firstly, a possibility density function (PDF) of the ranging errors is determined in the form of a dual-component GMM using Expectation-Maximum (EM) algorithm, which provides an approximation of the practical noise distribution in underground space. The PDF plays a crucial role in computing the upcoming overbounding PL. To ensure the mathematical tractability of the overbounding model, the bilateral boundaries of the errors are examined through the use of Cumulative Distribution Function (CDF). Next, in correspondence with the GMM-based PDF model, an asymmetrical CDF is obtained, necessitating the separate overbounding operation on both sides of the CDF. A heuristic adjustment on previous GMM-based PDF is conducted to increase the possibility density of the bilateral tail parts, ensuring sufficient but not excessive space to guarantee the validity of the PL. Initially, on the left side of the CDF, the weight and variance of the Gaussian component with a lower mean value in the GMM-based PDF are increased. Subsequently, the updated PDF will be shifted to the left to create a new version of the GMM-based PDF, with higher values in the left part of the CDF compared to the traditional one. Similar adjustment work needs to be conducted for the right part of the original CDF. After the adjustment operations on both sides, two different GMM-based PDFs can be obtained for every single base station, one is termed as left-boundary PDF while the other is right-boundary PDF. Finally, the predefined PDFs from different UWB base stations will be used to infer the overall PDF in position domain using a convolution operation. Based on this, PL can be computed from the inverse operation of the corresponding CDFs. [Results] In order to verify the methodology, experiments under practical simulated underground scenarios are conducted using six UWB base stations. Error models are constructed using sample data collected within a range of 3 to 93 meters. Evaluation of practical performance shows that the GMM-based bilateral bounding box reduces PL by more than 20% compared to traditional Gaussian-based calculations. [Conclusions] The GMM-based PDF can tighten the PL with a relatively low computational cost, enhancing the system usability.

[Objectives] The primary objective is to enhance the accuracy of vehicle trajectory prediction at intersections and address the challenges in predicting trajectories in multi-vehicle interaction scenarios. This is crucial for improving the safety and efficiency of autonomous driving and traffic management in complex urban intersections. [Methods] An Enhanced Adjacency Graph Convolutional Network-Transformer (EAG-GCN-T) vehicle trajectory prediction model is developed. The INTERACTION public dataset is employed, with data smoothing techniques applied to mitigate noise. Model comparison and validation experiments are conducted to assess performance. The model’s accuracy is evaluated by comparing error assessment indicators against different baseline models, analyzing interaction capabilities, generalization ability, and driving behavior recognition. The EAG-GCN-T model combines an Enhanced Adjacency Graph Convolutional Network (EAG-GCN) and a Transformer module. The EAG-GCN module accurately models spatial interactions between vehicles by considering relative speed and distance using an enhanced weighted adjacency matrix. The Transformer module captures temporal dependencies and generates future trajectories, improving spatiotemporal prediction ability. [Results] In long-term single-vehicle trajectory prediction, the Average Displacement Error (ADE) and Final Displacement Error (FDE) are reduced by 69.4%, 39.8%, and 33.3% and 71.9%, 32.5%, and 27.4% respectively, compared to CV, ARIMA, and CNN-LSTM models. In multi-vehicle interaction prediction, the FDE is reduced by 19.5% and 20.6% compared to the GRIP model. Compared with three interaction mechanisms, EAG-GCN-T achieves the lowest overall error across all time domains, with ADE/FDE values of 0.53 and 0.74, respectively. EAG-GCN-T achieves more reasonable Driving Area Compliance (DAC) and Trajectory Point Loss Rate (MR), demonstrating strong adaptability in ramps and roundabouts. The model accurately predicts driving behaviors such as following, lane-changing, evasion, and their impacts on trajectories, with predicted trajectories highly consistent with actual vehicle movements. [Conclusions] The EAG-GCN-T model effectively addresses vehicle trajectory prediction in multi-vehicle interaction scenarios at intersections. It demonstrates high accuracy, strong interactivity, and excellent generalization ability. This model provides a novel solution for vehicle trajectory prediction in intelligent transportation systems, offering significant potential for advancing autonomous driving and intelligent traffic management.

[Objectives] With accelerating urbanization and a surge in vehicle numbers, urban traffic systems face immense pressure. Intelligent transportation systems, a vital component of smart cities, are widely employed to improve urban traffic conditions, with traffic speed prediction being a key research focus. However, the complex coupling relationships and dynamically varying characteristics of urban traffic network nodes pose challenges for existing traffic speed prediction methods in accurately capturing dynamic spatio-temporal correlations. Spatio-temporal graph neural networks have proven to be among the most effective models for traffic speed prediction tasks. However, most methods heavily rely on prior knowledge, limiting the flexibility of spatial feature extraction and hindering the dynamic representation of road network topology. Recent approaches, such as adaptive adjacency matrix construction, address the limitations of static graphs. However, they often overlook the synergy between dynamic features and static topology, making it difficult to fully capture the complex fluctuations in traffic flow, which in turn limits prediction accuracy and adaptability. [Methods] To address these challenges, this study formulates urban traffic speed prediction as a multivariate time-series forecasting problem and proposes a traffic speed prediction model based on a Multivariate Time-series Dynamic Graph Neural Network (MTDGNN). Leveraging real-time traffic information and predefined static graph structures, the model adaptively generates dynamic traffic graphs to capture spatial dependencies through a graph learning layer and integrates them with static road network graphs to capture spatial dependencies from multiple perspectives. Meanwhile, the alternating use of graph convolution and temporal convolution modules constructs a multi-level spatial neighborhood and temporal receptive field, fully exploring the spatial and temporal features of traffic data. [Results] The MTDGNN model was tested on real traffic data from 397 road sections in eastern Beijing, collected between April 1, 2017, and May 31, 2017. Its prediction results were compared against nine benchmark models and seven ablation models. Compared to benchmark models, MTDGNN reduced the average MAE by at least 2.24% and the average RMSE by at least 3.98%. [Conclusions] Experimental results demonstrate that the MTDGNN model achieves superior prediction accuracy in MAE, RMSE, and MAPE evaluation metrics, highlighting its robustness and effectiveness in complex traffic scenarios.

[Objectives] Accurate and reliable traffic state prediction is essential for various applications in Intelligent Transportation Systems (ITS). However, the complexity of urban road networks makes it challenging to effectively model spatial dependencies between road segments, posing a significant obstacle to urban traffic forecasting. Traditional Graph Convolutional Networks (GCNs) are widely used for traffic prediction but fail to account for the unique characteristics of traffic networks, such as driving directions, turning rules, and varying spatial dependencies. This study aims to address these challenges by proposing a novel graph convolutional network model, the Turn-based Graph Convolutional Neural Network (TurnGCN), which better captures the complex spatial relationships in urban traffic networks. [Methods] TurnGCN models the urban road network as a heterogeneous graph, where edges represent turning relationships between road segments. Unlike traditional GCNs that rely on static adjacency matrices, TurnGCN introduces a turning table to label neighboring nodes and map their features into a structured Euclidean feature grid. A Convolutional Neural Network (CNN) is then applied to this grid to aggregate and fuse the spatial features of neighboring nodes. This approach allows TurnGCN to model the heterogeneity of turning relationships and learn their varying impacts on the central road segment. Additionally, the parameter-sharing nature of CNNs ensures that TurnGCN performs efficiently with relatively fewer trainable parameters. [Results] To validate the effectiveness of TurnGCN, extensive experiments were conducted on two real-world traffic datasets: Urban-150 from Seoul, South Korea, and SHSpeed from Shanghai, China. These datasets vary in sampling density and temporal resolution, presenting diverse evaluation challenges. The results demonstrate that TurnGCN consistently outperforms traditional GCNs and GCN variants enhanced with spatial attention mechanisms across multiple evaluation metrics. Specifically, TurnGCN excels in capturing heterogeneous spatial dependencies and modeling turning relationships in urban road networks. [Conclusions] TurnGCN provides a robust, efficient, and scalable solution for urban traffic prediction by explicitly modeling turn-based spatial relationships. It overcomes the limitations of traditional GCNs and attention-based models, achieving significant improvements in predictive performance while maintaining computational efficiency. These advantages highlight TurnGCN’s potential for practical applications in ITS, including traffic flow optimization, congestion management, and intelligent navigation systems.

[Objectives] Polycentric spatial strategies have emerged as effective solutions to urban challenges such as prolonged commuting, traffic congestion, air pollution, and environmental degradation. These strategies, widely implemented in Chinese cities, require robust evaluation to enhance the efficacy of urban spatial development policies. Traditional research approaches often combine quantitative and qualitative methods or develop evaluative indicators, yet they may lack explicit, measurable goals, and benchmarks, leading to reliance on empirical judgment. This can limit their practical application in directing the future development and investments for urban centers, subsequently affecting urban governance support. [Methods] To addressing these gaps, this study introduces a comprehensive technical framework that employs multiple machine learning models. Its primary aim is to delineate the optimal polycentric structure for a city by contrasting it with the present spatial distribution of employment centers. This comparative analysis enables a more empirical assessment of urban spatial strategy performance. Central to this framework is the goal of minimizing the city's average commuting distance, a metric significantly influenced by the urban spatial structure and commonly used in urban structure assessment. The study unfolds in several stages: initially, it utilizes a probabilistic mixture model to identify the urban polycentric structure and gauge the influence of individual centers. It then employs machine learning to investigate the non-linear interplay between urban employment centers, residential/occupational patterns, and commuting flows. Subsequently, an intelligent optimization algorithm calculates the optimal polycentric structure with the aim of minimizing commuting distances. The study culminates in the development of performance evaluation indices that quantitatively measure the urban spatial structure's current state against the ideal, highlighting avenues for future optimization. [Results] Empirical evaluation is conducted across the central urban zones of thirteen prefecture-level cities within the Beijing-Tianjin-Hebei metropolitan region, yielding three principal findings. First, each city exhibits a distinct polycentric structure, with primary employment centers situated centrally within municipal districts and secondary centers dispersed in the periphery. Second, the optimization results indicate a general trend of decentralization, with a movement from city centers to the periphery. Third, cluster analysis categorizes the cities into four groups, with strategic recommendations provided for each group, largely aligning with existing local government urban and spatial plans. [Conclusions] This study's contributions are multifaceted. It proposes a methodology for ascertaining the optimal polycentric pattern, thus creating a standard for performance evaluation and enabling quantifiable assessments. It offers a comprehensive method for assessing the execution of urban polycentric spatial strategies and evaluating the intricacy of the city's polycentric structure. Finally, the application of these methods in the Beijing-Tianjin-Hebei region provides analytical outcomes that are instrumental for the future urban refinement governance of the area.

[Objectives] City-wide traffic flow prediction plays a crucial role in intelligent transportation systems. Traditional studies partition road networks into grids, represent them as graph structures with grids as nodes, and use graph neural networks for region-level prediction. However, this region-based approach overlooks the relationships between individual roads, making it difficult to reflect traffic flow changes of roads. Methods based on road segment data can better capture spatial connections between roads and enable more accurate traffic flow predictions. However, mapping trajectory data to roads presents challenges such as redundant data and trajectory mismatches, and traffic flow data after mapping is sparse. Existing methods struggle to effectively capture the spatial correlation in sparse traffic conditions. [Methods] To address these issues, this study proposes an Attention Spatio-Temporal Neural Network (ASTNN) model for road-level sparse traffic flow prediction. The model first preprocesses trajectory data and applies Hidden Markov Model (HMM)-based map matching to obtain road-level traffic flow data. It then introduces an adaptive compact 2D image representation method to model the road network as a 2D image, where road segments are represented as pixel points. Based on an analysis of the spatial and temporal characteristics of traffic flow, two new attentional spatiotemporal blocks are proposed: Attentional Spatio-Temporal Memory Block (ASTM block) for mining temporal correlations and attentional spatial-temporal focusing block (ASTF block) for extracting spatial sparse features. By integrating these two blocks with external information, ASTNN is constructed to achieve road-level traffic flow prediction. [Results] This study uses Chengdu taxi trajectory data as a case study. After preprocessing trajectory data and mapping traffic flow, the proposed model is validated on a five-level road network within Chengdu’s third ring area. Results indicate that the proposed data processing method reduces trajectory-to-road network matching time by 73.6%. In the comparative experiments with existing models, such as Convolutional Neural Network (CNN), Convolutional Long Short-Term Memory (ConvLSTM), Gated Recurrent Unit (GRU), and Spatial-Temporal Neural Network (STNN), ASTNN achieves the highest prediction accuracy in terms of Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and R-squared (R²). Furthermore, the study confirms the significant improvement in prediction accuracy when incorporating temperature data into ASTNN, providing new insights for optimizing model performance. [Conclusions] The ASTNN model proposed in this study provides an effective framework for city-wide, road-level sparse traffic flow prediction, offering valuable insights for intelligent transportation systems.