Predicting rail transit passenger flow is crucial for modifying the metro schedule. To increase prediction accuracy, a model is proposed that combines long short-term memory (LSTM) with single spectrum analysis (SSA). Firstly, a stepwise decomposition sampling (SDS) strategy based on SSA progressive decomposition is proposed as a solution to the data leaking issue in traditional sequence decomposition. Then, based on this strategy, the passenger flow time series with complex features is decomposed into a relatively single trend and fluctuation component. Finally, the LSTM network is employed to perform short-term predictions on each component separately. The predicted value of each component is accumulated to obtain the original passenger flow’ predicted result. The example shows that, compared with the single LSTM and other hybrid models, the proposed method offers a greater overall prediction accuracy in the experimental days, and the method has specific applicability.

In the cargo transportation system, the hub-and-spoke transport network can make full use of the scale effect between logistics hubs and reduce logistics costs. Joint transportation of multiple modes of transportation can give full play to the advantages of different modes of transportation, which not only reduces logistics costs but also improves transportation efficiency. Therefore, this paper combines the advantages of multimodal transportation and hub-and-spoke network, and establishes an optimization model of multimodal hub-and-spoke transport network under demand uncertainty. The model takes into account hub capacity constraints and customer satisfaction with respect to transportation time, and to facilitate the model solution, we utilize the fuzzy expected value method and the fuzzy chance constraints based on credibility to clarify the uncertain variables in the model. We use mixed coding to describe the selection of hubs, assignment of nodes, and choice of transportation modes in this study and use the NSGA-II algorithm with local reinforcement to solve the problem. Finally, numerical experiments are designed to verify the validity of the model and algorithm through sensitivity analysis of relevant parameters, determine the reasonable number of hubs and confidence level, and obtain the influence of the change of hub capacity limit and the ratio of single and double hub transit on the research objectives. The results show that: the NSGA-II algorithm with local reinforcement can significantly improve the convergence speed of the algorithm; There is benefit inversion between economic cost and time cost, and the pursuit of economic cost minimization and time cost minimization, respectively, will lead to different choices of the number of hubs; Increasing the ratio of goods transfer between hubs is beneficial for fully utilizing the scale effect between hubs, achieving the goal of reducing economic costs, but at the same time, it will increase time costs.

The rapid growth in the number of electric bicycles (e-bicycles) has greatly improved daily commuting for residents, but it has also increased traffic collisions involving e-bicycles. This study aims to develop an autonomous emergency braking (AEB) system for e-bicycles to reduce rear-end collisions. A framework for the AEB system composed of the risk recognition function and collision avoidance function was designed, and an e-bicycle following model was established. Then, numerical simulations were conducted in multiple scenarios to evaluate the effectiveness of the AEB system under different riding conditions. The results showed that the probability and severity of rear-end collisions involving e-bicycles significantly decreased with the application of the AEB system, and the number of rear-end collisions resulted in a 68.0% reduction. To more effectively prevent rear-end collisions, a low control delay (delay time) and suitable risk judgment criteria (TTC threshold) for the AEB system were required. The study findings suggested that when a delay time was less than or equal to 0.1 s and the TTC threshold was set at 2 s, rear-end collisions could be more effectively prevented while minimizing false alarms in the AEB system. Additionally, as the deceleration rate increased from 1.5 m/s² to 4.5 m/s², the probability and average severity of rear-end collisions also increased by 196.5% and 42.9%, respectively. This study can provide theoretical implications for the design of the AEB system for e-bicycles. The established e-bicycle following model serves as a reference for the microscopic simulation of e-bicycles.

This study mainly investigates the maximum length of CAV (Connected and Autonomous Vehicle) platoons in a heterogeneous traffic flow environment. By employing MATLAB to simulate the heterogeneous traffic flow on expressways, this study focuses on the maximum platoon length of CAV platoons and explores their impact on the traffic flow characteristics on expressways. Firstly, based on four different car-following modes of heterogeneous traffic flow, F-STCA (Flexible–Symmetric Two-Lane Cellular Automata Model) and the symmetric lane-changing strategy, the study refines and improves the construction of the NaSch (Nagel and Schreckenberg) model introduced into the Gipps safety distance formula. The whole improvement process is based on the acceleration decay characteristics of vehicles on expressway. Secondly, the congestion situations under different maximum platoon lengths are compared using fundamental heat maps of the simulation data. The evolution of the fundamental diagram with changes in maximum platoon length is studied to investigate the impact and magnitude of maximum platoon lengths on the road capacity under different CAV permeabilities. Finally, the study explores the stability and safety of heterogeneous traffic flow involving CAV platoons using SD (Standard Deviation) and TERCRI (Time-Exposed Rear-End Collision Risk Index). The results show that when the CAV’s permeability does not reach a high level on heavily trafficked sections of the expressway, considering the overall average speed, the maximum platoon length should not be set too long and should be around five vehicles. This restriction does not apply when the CAV permeability approaches 100%.

To lighten the workload of train drivers and enhance railway transportation safety, a novel and intelligent method for railway turnout identification is investigated based on semantic segmentation. More specifically, a railway turnout scene perception (RTSP) dataset is constructed and annotated manually in this paper, wherein the innovative concept of side rails is introduced as part of the labeling process. After that, based on the work of Deeplabv3+, combined with a lightweight design and an attention mechanism, a railway turnout identification network (RTINet) is proposed. Firstly, in consideration of the need for rapid response in the deployment of the identification model on high-speed trains, this paper selects the MobileNetV2 network, renowned for its suitability for lightweight deployment, as the backbone of the RTINet model. Secondly, to reduce the computational load of the model while ensuring accuracy, depth-separable convolutions are employed to replace the standard convolutions within the network architecture. Thirdly, the bottleneck attention module (BAM) is integrated into the model to enhance position and feature information perception, bolster the robustness and quality of the segmentation masks generated, and ensure that the outcomes are characterized by precision and reliability. Finally, to address the issue of foreground and background imbalance in turnout recognition, the Dice loss function is incorporated into the network training procedure. Both the quantitative and qualitative experimental results demonstrate that the proposed method is feasible for railway turnout identification, and it outperformed the compared baseline models. In particular, the RTINet was able to achieve a remarkable mIoU of 85.94%, coupled with an inference speed of 78 fps on the customized dataset. Furthermore, the effectiveness of each optimized component of the proposed RTINet is verified by an additional ablation study.

Subway station-level peak hour ridership (SPR) is a crucial input parameter for multiple applications, including the planning, design, construction, and operation of stations. However, traditional SPR estimation techniques may produce biased results. A unified peak hour factor (PHF) extracted from the line level is generally set for all attributed stations, which ignores the possible peak deviation that arises between the station and line and the wide variation of PHFs in practice. This study presents a comprehensive and refined estimation framework for SPR that accommodates the peak deviation context by introducing the peak deviation coefficient (PDC). Moreover, the estimation of the PDC and PHF variability is improved by constructing spatial regression based relationship models. The empirical results show that the proposed approach exhibits wider applicability and a higher prediction precision across all types of peak periods considered as compared to conventional methods (i.e., MAPE decreases of 0.115–0.351). The findings demonstrate the importance of the consideration of the peak deviation scenario and the spatial dependency in SPR estimation to achieve better decision making. Moreover, the underlying influencing mechanism of the PHF and PDC at distinct peak periods is further revealed using the spatial model. This provides critical theoretical references and policy implications to prudently deploy land-use resources to balance the travel demand between peak and off-peak periods and thus enhance the line operation efficiency.