Current Issue

2026, Volume 44,  Issue 4

  Previous Issue
Display Method:
2026, 44(4): .
Abstract(33) HTML (22) PDF(1)
Abstract:
Research Progress and Trends of Integrated Passenger–Freight Transportation in Urban Rail Transit
HE Shanglu, LIU Mengdie, YE Mao, ZHENG Tianlin
2026, 44(4): 1-12. doi: 10.3963/j.issn.1674-4861.2026.04.001
Abstract(55) HTML (25) PDF(5)
Abstract:
To systematically review the research progress in integrated passenger-freight transportation in urban rail transit, 141 publications were used as the sample. The publications were retrieved from the China National Knowledge Infrastructure (CNKI), Wanfang Data, and Web of Science between January 2003 and May 2026. Statistical analysis of the literature, keyword co-occurrence analysis, and content synthesis were combined. Research hotspots, practical characteristics, and existing problems were identified, together with a comparative analysis of representative practices in China and abroad. Research results show that research in this field has grown rapidly since 2018, with metro systems as the main research object. Research hotspots center on network facility location and routing optimization and system operation scheduling. The former has expanded from separate facility location or routing optimization to integrated location-allocation-routing decisions. The latter has expanded from capacity allocation under a fixed passenger timetable to coordinated optimization of train formation, timetabling, and freight-flow allocation. Operations research methods remain dominant. Simulation is mainly used for parameter es-timation and plan validation. Machine learning has seen limited application and is used primarily for demand processing and auxiliary assessment. Current research has several limitations. Freight demand lacks empirical calibration, and the relationships between operational capacity and planning and scheduling decisions have not been adequately examined. Task adjustment under disruptions and machine-learning-assisted solution methods remain underexplored. Facilities and equipment lack validation under actual operating conditions. Safety standards specific to integrated passenger-freight transportation have yet to be established. Mechanisms for multi-actor coordination and intermodal connections remain incomplete. Operating practices have gradually shifted from early dedicated systems to the reuse of existing lines. Recent pilot projects in China mainly employ single-line freight hitching services or shared passenger-freight carriage modes. Their sustained operation depends primarily on stable freight demand, economic viability, safety boundaries, and policy support. Future research should focus on network planning based on empirically calibrated demand and accounting for operational capacity. Dynamic scheduling of freight tasks and intelligent solution methods under disruptions should be developed. Standards for cargo admission, transport safety, and facility retrofitting should be improved. Integration of rail transport with ground and low-altitude delivery and multi-actor collaborative governance should be strengthened.
Review of Low-Altitude Route Planning Methods Considering Airspace Environment
ZHOU Zhiyi, GUO Junhua, YAN Lixin
2026, 44(4): 13-23. doi: 10.3963/j.issn.1674-4861.2026.04.002
Abstract(52) HTML (15) PDF(2)
Abstract:
The growth of low-altitude flight activities and the expansion of application scenarios place increasing demands on the safety, environmental adaptability, and coordination of multiple aircraft in route planning. Focusing on low-altitude route planning in complex airspace environments, this review systematically summarizes research progress from three perspectives: task evolution, influencing factors, and planning algorithms, so as to inform route design and the specification of operating conditions. According to the expansion of planning objects, the evolution of low-altitude route planning is divided into four stages, i.e., military penetration, civil single-aircraft, multi-aircraft coordination, and infrastructure networking, and the core constraints and optimization objectives of each stage are summarized. Regarding influencing factors, ground environments, airspace management, flight demand, and socio-ecological conditions are examined to summarize their modeling approaches, and their effects are divided intoThe growth of low-altitude flight activities and the expansion of application scenarios place increasing demands on the safety, environmental adaptability, and coordination of multiple aircraft in route planning. Focusing on low-altitude route planning in complex airspace environments, this review systematically summarizes research progress from three perspectives: task evolution, influencing factors, and planning algorithms, so as to inform route design and the specification of operating conditions. According to the expansion of planning objects, the evolution of low-altitude route planning is divided into four stages, i.e., military penetration, civil single-aircraft, multi-aircraft coordination, and infrastructure networking, and the core constraints and optimization objectives of each stage are summarized. Regarding influencing factors, ground environments, airspace management, flight demand, and socio-ecological conditions are examined to summarize their modeling approaches, and their effects are divided into constraints on route eligibility and cost factors affecting scheme comparison. Regarding planning algorithms: ①To improve trajectory feasibility in complex environments, potential-field guidance, adaptive sampling, and kinematic constraints are incorporated to improve path search and segment connection, while integrated planning and tracking and navigation uncertainty modeling bring trajectory execution and collision risk into the planning process. ②To address safety and efficiency under changing weather conditions, terrain and wind-field modeling, icing-risk constraints, and distributionally robust optimization are used to reduce flight costs, identify feasible trajectories, and mitigate delivery delays, respectively. ③To support multiple aircraft sharing airspace, integrated network design and demand modeling, quasi-dynamic traffic assignment, and combined strategic and tactical conflict management extend planning from individual missions to route networks and the allocation of spatial and temporal resources. The re sults indicate that existing planning algorithms have advanced from geometric obstacle avoidance to kinematic adap tation, navigation risk assessment, and network traffic assignment; however, management rules can hardly enter planning models directly, environmental indicators have not yet been converted into tiered route-use standards, and algorithm validation mostly remains at the level of single missions, which restricts the sustainable operation of public route networks. In the future, research on safety margins and operating conditions of heterogeneous aircraft should be conducted, and route adjustment and mission rescheduling under weather forecast uncertainty should be improved to support safe route use; capacity allocation incorporating feedback on trajectory execution deviations and alternative corridor design should also be investigated to balance operational efficiency under normal conditions with recovery capability following disruptions.
A Frontal Automotive Airbag Ignition Decision Model Based on Weakly Supervised Timing-Aware Learning
CHEN Chen, LU Weitao, LI Bin, HUANG Dongzi, XIONG Zhao, QIAN Chuang, CHEN Yanshuo
2026, 44(4): 24-36. doi: 10.3963/j.issn.1674-4861.2026.04.003
Abstract(31) HTML (17) PDF(1)
Abstract:
To address the difficulty of simultaneously ensuring trigger reliability and ignition-timing accuracy for frontal automotive airbags when full-vehicle crash data are expensive, positive samples are limited, and frame-level labels are unavailable, this study investigates a weakly supervised timing-aware decision method and develops a corresponding airbag ignition model. First, an optimal ignition-timing matrix for frontal crash conditions is determined from dummy head trajectories and airbag deployment characteristics measured in full-vehicle crash tests. Multichannel longitudinal body-acceleration signals are then processed by zero-phase filtering, time-origin normalization, channel alignment, and standardization. Pointwise weak labels are generated using a rule that combines an amplitude threshold, multichannel agreement, and a minimum-duration constraint. Window-level samples are constructed with a 15 ms window and a 5 ms stride, and Gaussian timing-aware weights centered on the target ignition time direct the model toward the millisecond-level critical trigger region. A two-layer Bi-LSTM extracts bidirectional temporal features, while inverse-frequency class weighting, negative-window downsampling, and amplitude-scaling and noise augmentation that preserve relative timing semantics mitigate small-sample and class-imbalance effects. During inference, the output probability sequence is coordinated with condition-specific thresholds, consecutive-window requirements, and delay parameters to produce a calibratable, interpretable, and traceable ignition strategy. Evaluation on an independent test set shows a 100% trigger rate for six frontal crash conditions and a 0% false-trigger rate for negative conditions, including misuse, low-speed, and side-impact cases. The macro-averaged absolute timing error (SMAE) is 0.625 ms, and both R5 and R10 reach 98.81%. The mean absolute errors are 0 ms at 23, 30, 32, 35, 50 km/h and 3.75 ms at 56 km/h. Ablation results further show that removing timing-aware weighting or data augmentation, or introducing excessive temporal perturbation, markedly increases ignition-timing error; under the present small-sample weak-supervision setting, the Bi-LSTM backbone also outperforms bidirectional GRU and Temporal CNN in timing alignment. These results demonstrate that combining the optimal timing matrix, weak-label generation, timing-aware weighting, and explicit strategy coordination unifies the objectives of whether and when to trigger, enabling millisecond-level timing alignment while strictly suppressing negative-condition triggers.
Functionality Resilience of Autonomous Vehicles: Case Studies of Lateral Control and Positioning
FANG Yukun, MIN Haigen, HU Gongda, ZHAO Xiangmo
2026, 44(4): 37-47. doi: 10.3963/j.issn.1674-4861.2026.04.004
Abstract(26) HTML (14) PDF(1)
Abstract:
Autonomous vehicles (AVs) face functionality degradation and safety risks in complex dynamic environments. Such risks arise from uncertainties, including the coupling of internal subsystems, changes in the external environment, and interactions among multiple traffic participants. To address this problem, a functionality-centered resilience framework together with its characterization, quantification, and enhancement methods is proposed. Functionality resilience is formalized as a triplet (A, F, U) consisting of an adverse event (A), a functionality (F), and uncertainty (U). Based on the geometric characteristics of the performance curve Q(t), functionality resilience is evaluated from both global and local dimensions. The global resilience loss RL quantifies the functionality loss during adverse events, while the local geometric features of the performance curve characterize the absorption and recovery behavior of the system. Resilience enhancement principles centered on reducing epistemic uncertainty, optimizing the functionality level, and introducing performance feedback are summarized. In the vehicle lateral control case, the functionality degradation of a pure pursuit controller in sharp-curve scenarios is addressed. An adaptive look-ahead distance method that fuses real-time vehicle speed and path curvature is designed, and a closed-loop correction consisting of proportional-integral-moving-average-derivative feedback of the lateral error is introduced. Vehicle tests and simulation experiments show that, for the proposed strategy, the resilience loss RL at two successive sharp curves decreases from 63.93 to 49.50 and from 61.50 to 45.58. The absolute value of the performance drop L1 decreases from 6.73 to 5.15 and from 5.83 to 3.94. The recovery speed k2 increases from 0.530 to 0.683 and from 0.565 to 0.864. In the vehicle positioning case, a position"prediction-correction"network is developed. Global navigation satellite system (GNSS) positioning errors are used as supervisory information to train the correction network, thereby improving the positioning functionality resilience of the integrated navigation system during GNSS outages. Under three GNSS outage scenarios, the positioning functionality resilience of the prediction-correction strategy is superior to that of dead reckoning and the " $p-\Delta p$ " architecture. The resilience loss RL at the second outage decreases from 420.13 for dead reckoning to 141.41. The results indicate that the (A, F, U) framework can effectively characterize, quantify, and enhance the functionality resilience of AVs. Reducing epistemic uncertainty and optimizing the functionality level are effective ways to improve functionality resilience.
Research on a Safety Dynamic Early Warning Model for Hazardous Chemical Tankers under Multi-Parameter Spatio-Temporal Conditions
LIAO Zhengdong, ZHOU Shouqin, WANG Can, LEI Gaopan, LIU Yi, LU Dongyan
2026, 44(4): 48-57. doi: 10.3963/j.issn.1674-4861.2026.04.005
Abstract(39) HTML (26) PDF(0)
Abstract:
Existing early-warning systems for tank trucks transporting hazardous chemicals commonly rely on fixed thresholds and fail to characterize the dynamic coupling among multi-source spatio-temporal parameters, which leads to false and missed alarms. To address this limitation, a spatio-temporal multi-parameter dynamic safety ear ly-warning model was proposed based on a long short-term memory autoencoder and the multivariate state estimation technique (LSTM-AE-MSET). The model integrates multi-source sensor data with spatio-temporal attributes, including temperature, pressure, liquid level, inclination, and vibration, and incorporates road spatial information acquired via the global navigation satellite system (GNSS), the inertial measurement unit (IMU), and map application programming interfaces (API). The LSTM-AE extracts temporal features, captures long-term dependencies among variables, and maps high-dimensional inputs into a low-dimensional latent space; an MSET memory matrix is con-structed based on historical healthy-state data, and the early-warning thresholds are dynamically adjusted through state estimation and residual analysis. The Tennessee Eastman Process (TEP) dataset is used to evaluate the model. The dataset comprises 41 process measurement variables, 12 manipulated variables, and 21 fault modes, covering typical anomalies such as step disturbances, random variations, and valve stiction. After being trained on the TEP dataset, the model is transferred to a tank-truck operating scenario to identify anomalies such as abrupt temperature changes, pressure drift, and liquid-level oscillations under different spatio-temporal conditions. The results show that the model achieves an average coefficient of determination (R2) of 0.621 8 and a mean squared error of 4.301 4, with medians of 0.998 5 and 0.000 1, respectively, indicating excellent fitting performance under most normal operating conditions. Compared with the standalone MSET method, the proposed model increases the anomaly-detection recall from 57% to 88% and the F1-score from 60% to 74%, improves the overall accuracy by 2 percentage points, and reduces the missed-alarm rate by 72%. The proposed method enhances the identification of abnormal op erating states in hazardous-chemical tank trucks and provides technical support for real-time dynamic early warning under complex operating conditions.
Risk Factor Analysis and Path Identification Method for Runway Excursion During Landing Based on Bayesian Network
QI Xinge, YANG Ruitong, LEI Bowen, LIU Chang
2026, 44(4): 58-67. doi: 10.3963/j.issn.1674-4861.2026.04.006
Abstract(26) HTML (15) PDF(1)
Abstract:
To address the complex relationships among risk factors of runway overrun and veer-off during landing, the difficulty of directly reflecting flight-physical relationships through statistical associations, and the lack of quantitative diagnosis of key risk paths, a runway excursion risk analysis and path identification method integrating association rule mining and Bayesian networks (BN) was investigated. Based on 224 valid landing records obtained from the quick access recorders (QAR) of a B737-800 fleet from an airline, 15 key parameters in three categories, including aircraft states, pilot operations, and environmental wind conditions, were selected. The Apriori algorithm was used to mine strong associations among the parameters and generate candidate links for the BN. The directions of these links were then verified according to the energy-dissipation mechanism of runway overrun and the lateral-dynamics mechanism of runway veer-off, and statistical associations lacking a physical basis were removed, thereby establishing a network structure jointly constrained by data associations and flight-physical mechanisms. The expectation-maximization (EM) algorithm was employed for network parameter learning, and mutual information (MI) was used to identify key risk factors. Bayesian backward inference combined with the ratio of variation (ROV) was further applied to quantitatively diagnose upstream association paths under high-risk target states. The results show that when ground speed was in a high-risk state, the ROV values of the paths associated with pitch angle and brake pressure were 0.348 and 0.306, respectively; when vertical load was in a high-risk state, the ROV value of the path associated with descent rate was 3.911; for runway veer-off, when bank angle was in a high-risk state, the ROV value of the path associated with rudder position was 0.222. The quantitative results distinguish key upstream association paths under different high-risk target states, extending runway excursion risk analysis from key risk factor identification to quantitative diagnosis of key risk paths.
Comprehensive Ability Evaluation Method for Autonomous Vehicle Test Safety Officers
MIAO Mingyue, PEI Shunsi, GU Liyuan, ZHAO Fangxia, HUANG Lihua, LAI Jianhui
2026, 44(4): 68-78. doi: 10.3963/j.issn.1674-4861.2026.04.007
Abstract(25) HTML (16) PDF(0)
Abstract:
To address the absence of a quantifiable and operational evaluation framework for the comprehensive competence of autonomous vehicle test safety officers, and the inability of existing driver assessment systems to cover the full"monitor-takeover-feedback"responsibility chain, a three-dimensional evaluation index system encompassing reaction ability, driving ability, and description ability was constructed based on the job characteristics of test safety officers.Fifty frontline test safety officers participated in a staged experiment covering three modes: no-driving task, manual driving, and autonomous driving.Six categories comprising thirteen raw data indicators were collected, including reaction time, driving operations, and problem descriptions.For reaction ability evaluation, reaction time mean, variance, and error count were adopted as core indicators.Six algorithmic combinations involving Gaussian Mixture Model (GMM), Extremely Randomized Trees (ET), Random Forest (RF), and eXtreme Gradient Boosting (XGBoost) were preliminarily compared and screened through a four-level progressive strategy of clustering quality, information retention, variable utilization sufficiency, and classification accuracy, based on the silhouette coefficient (0.82), Akaike information criterion (AIC) of 186.3, weight distribution uniformity, and PRF (Precision, Recall, $F_{1}$-score).The GMM+ET fusion model was identified as the optimal evaluation model, achieving precision, recall, and $F_{1}$-score all equal to 1.00, with a mean accuracy of 0.99 under ten-fold cross-validation. For driving ability evaluation, a multiple linear regression model (adjusted $R^{2}=0.552, \operatorname{Prob}$ (F-statistic) $<0.01$) with accident frequency as the dependent variable was constructed, identifying maximum braking value, average throttle input, and optimal speed deviation as the three core influencing indicators.For description ability evaluation, a multi-expert grading method was employed to quantify the accuracy and conciseness of safety officers'problem feedback.Based on accident attribution statistics and Delphi method expert calibration, the weights of reaction ability, driving ability, and description ability were determined as 0.402, 0.496, and 0.102, respectively.The results demonstrate that the three ability dimensions are mutually independent, with a Pearson correlation coefficient of-0.123 between reaction and driving abilities.Driving ability score exhibits the strongest correlation with comprehensive ranking $(r=0.87)$, representing the core factor influencing overall competence.With a ten-fold cross-validation mean accuracy of 0.99 and $F_{1}$-score of 1.00, the GMM+ET fusion model significantly outperforms GMM+RF (mean $F_{1}$-score of 0.63) and k-means+XGBoost (mean $F_{1}$-score of 0.58) in variable utilization adequacy and evaluation accuracy.
AIS-Information-Aided Ship Association and Identification in SAR Images
FAN Liang, XU Zidong, ZHANG Di, SUN Yongqiang, ZHU Xiaodong
2026, 44(4): 79-89. doi: 10.3963/j.issn.1674-4861.2026.04.008
Abstract(110) HTML (23) PDF(0)
Abstract:
To address the lack of identity attributes for ship targets in synthetic aperture radar (SAR) imagery and the heterogeneous temporal, spatial, and geometric errors between SAR imagery and automatic identification system (AIS) data, a multi-dimensional feature-based stepwise association method is proposed for ship identity identification in SAR imagery. The method is designed to mitigate temporal asynchrony, ship-shape deviation, and locally non-uniform spatial distortion during SAR AIS data fusion. AIS observations within 30 min before and after SAR imaging are used to construct continuous ship trajectories through cubic spline interpolation, and the ship positions corresponding to the SAR imaging time are reconstructed to reduce motion-induced temporal matching errors. An instance segmentation network is used to extract pixel-level ship masks and determine target positions from mask centroids. To address local protrusions and asymmetric shapes caused by side-looking imaging, sidelobe effects, and strong scattering from ship superstructures, principal component analysis (PCA) is employed to estimate the ship principal axis, and an iterative principal-axis correction method is developed to reduce the influence of asymmetric protrusion regions on orientation estimation. The corrected ship principal axis, length, and width are then used as geometric features. After SAR and AIS targets are transformed into a common geographic coordinate system, iterative closest point (ICP) is applied for global coarse registration. K-means is subsequently used to partition the scene into locally distorted regions according to the spatial distribution of targets, while the product of ship length and width is introduced as a scale feature for regional ICP matching. Accurate association is finally performed in a joint position scale feature space. Experiments are conducted using real SAR imagery and AIS data collected in the Wusongkou and Yangtze River estuary waters of Shanghai, including 72 463 AIS records from 1 171 vessels. The instance segmentation network achieves a precision of 0.874, a recall of 0.727, and a mean average precision at an intersection over union threshold of 0.5 (mAP50) of 0.821. The average distance error between SAR and AIS targets decreases from 2.469×10-3(°) before calibration to 0.368×10-3(°) after applying the proposed method, corresponding to reductions of 74.7%, 77.2%, and 77.8% compared with conventional ICP, nearest-neighbor matching, and the Hungarian algorithm, respectively. The results demonstrate that the proposed method can jointly exploit AIS trajectory information, SAR target positions, and ship-scale features while mitigating the effects of AIS temporal asynchrony, SAR ship-shape deviation, and locally non-uniform geometric distortion over large-area imagery. The method is particularly suitable for complex waters characterized by dense ship traffic, closely spaced targets, and spatially varying positioning errors, and can provide technical support for ship identity assignment in SAR imagery, maritime traffic supervision, abnormal-target screening, and multi-source maritime situational awareness in emergency search and rescue.
A Vertiport Location Method for Urban Low-Altitude Air-Ground Coordinated Transportation
HOU Jingjia, SUN Siyuan, WANG Bingjie, YANG Jianting, GUO Jianbo
2026, 44(4): 90-101. doi: 10.3963/j.issn.1674-4861.2026.04.009
Abstract(30) HTML (15) PDF(1)
Abstract:
Against the backdrop of the rapid development of the low-altitude economy, urban air mobility (UAM), which employs electric vertical take-off and landing (eVTOL) aircraft as the carrier, has emerged as a new pathway to breaking through urban traffic bottlenecks. Vertiport siting is the key to achieving efficient air-ground coordination between low-altitude and ground transportation. Existing studies neglect the probability distribution of users'travel willingness and fail to quantify the impact of air traffic on ground road network flow allocation during the dynamic air-ground interaction. To address this gap, an air-ground parallel operation mode of “non-motorized access-egress + eVTOL” is proposed, and a travel willingness probability model for the urban air-ground coordinated transportation network is developed, based on the multinomial logit (MNL) model, to characterize users' choice behavior between pure ground travel and low-altitude travel. A bilevel programming model for vertiport siting is then established, with the upper layer devoted to location decision-making and the lower layer to user equilibrium assignment; the ground impedance is calculated by the Bureau of Public Roads (BPR) function, and the interaction mechanism between air and ground traffic is quantified through an iterative coupled solution of“ground impedance-mode choice probability-demand reallocation.”Numerical results show that an optimal number of vertiports exists in the air-ground coordinated network: in the case study, the operating benefit reaches its peak of 87 200.74 yuan when the number of vertiports B is 6, and the network traffic impedance is improved by nearly 14% relative to the pure ground network. The resulting low-altitude network reshapes users' travel mode choices and origin-destination (OD) demand allocation by diverting flow from critical links, thereby forming an air-ground coupling mechanism of “demand shift-flow optimization-efficiency improvement.”It is further found that once the unit eVTOL fare exceeds 7 yuan, both users' low-altitude travel willingness and the mitigating effect of air traffic on ground impedance decline significantly.
Collaborative Optimization of Continuous Berth Allocation and Shore Power Scheduling via a Q-Learning-Based Selection Hyper-Heuristic
Ya rigui, ZHANG Yu, LU Xuan, TANG Kexin
2026, 44(4): 102-115. doi: 10.3963/j.issn.1674-4861.2026.04.010
Abstract(27) HTML (15) PDF(0)
Abstract:
This study addresses the joint scheduling of continuous berth allocation and shore power connection under uncertainty in ships' power demand. A mixed-integer programming model is formulated to minimize the total cost. The model considers non-overlapping spatiotemporal berth occupation, shore power coverage, ship-shore interface compatibility, and shore power capacity constraints. An uncertainty budget is introduced to reformulate the capacity constraint into a linear robust counterpart. A Q-learning-based selection hyper-heuristic algorithm (QLSHH) is proposed. A hybrid encoding scheme represents berthing positions and shore power decisions. Q-learning dynamically selects low-level heuristic operators, and eight complementary operators are integrated with a structured decoding mechanism. For small-scale instances, the average deviation of QLSHH from the exact solutions of Gurobi ranges from 0.1% to 5.0%. For medium and large-scale instances, QLSHH reduces the total cost by 4.12%-11.65% compared with a genetic algorithm (GA). It outperforms an adaptive large neighborhood search algorithm (ALNS) by 3.75%-25.33%. The average shore power utilization rate remains between 90% and 95.41%. As the uncertainty budget increases, the shore power utilization rate decreases and the total cost increases. A higher shore power capacity limit improves the utilization rate and reduces the total cost. However, the marginal benefit diminishes when the capacity exceeds approximately 4 500 kW. Therefore, the capacity should be configured according to practical electricity demand under a given uncertainty budget.
Research on a UAV Task Allocation and Path Planning Model Based on Dynamic Payload
WANG Lili, MA Ziyang
2026, 44(4): 116-126. doi: 10.3963/j.issn.1674-4861.2026.04.011
Abstract(30) HTML (17) PDF(0)
Abstract:
To capture the effects of payload reduction on energy estimation and route feasibility in multi-stop deliv ery, a joint UAV task allocation and path planning model is developed. The model considers a single-depot scenario and minimizes fixed UAV utilization, travel, energy, and time-window violation costs. The problem is formulated as a capacitated vehicle routing problem with time windows (CVRPTW), in which time-window violations incur additional costs. The remaining payload is updated for each flight segment according to the delivery-station visitation sequence. UAV operations are divided into vertical takeoff, horizontal flight, and vertical landing. A phase-based dynamic-payload energy model accounts for altitude-dependent air density and the cumulative effects of repeated takeoffs and landings. The resulting route energy cost is incorporated into the total economic cost. A two-stage solution method is designed. In the first stage, an improved A* algorithm searches for the shortest obstacle-avoiding paths between stations. Directional search and synchronous bidirectional expansion reduce redundant searches, while greedy pruning is applied to construct the distance matrix. In the second stage, an improved Clarke-Wright savings algorithm (C-WSA) merges the initial routes based on this matrix. The distance-based savings metric is extended to a composite economic savings metric comprising the four cost components. Simulation results show that the model generates seven feasible delivery routes. Takeoff and landing energy accounts for 22.53%—30.20% of the total energy consumption of each route. Removing the energy constraint reduces total travel distance by 10.92% but produces infeasible routes whose energy consumption exceeds the battery capacity. For the same delivery routes, total energy estimates from equivalent constant-payload and segment-wise dynamic-payload calculations differ by 16.38%. This difference indicates that dynamic-payload effects cannot be neglected in energy estimation. The improved A* algorithm reduces search time by 73.57% relative to the conventional A* algorithm. The improved C-WSA reduces total economic cost by 26.38%—36.64% relative to the benchmark algorithms. These results demonstrate the effectiveness of the proposed two-stage solution method.
Integrated Planning Model of Fixed-Route and Demand-Responsive Transit Services under Rail Feeder Perspective
LU Haiyang, NING Jiang, CHEN Xuewu
2026, 44(4): 127-137. doi: 10.3963/j.issn.1674-4861.2026.04.012
Abstract(23) HTML (13) PDF(0)
Abstract:
To address the mismatch between supply and demand in conventional fixed-route bus services caused by significant peak and off-peak variations in rail feeder demand in residential areas, an integrated service planning model combining fixed-route and demand-responsive transit is developed. Taking the number of operating vehicles and service frequency as decision variables, a two-stage integrated service planning model is formulated to account for the differences in peak and off-peak passenger demand. During peak periods, the existing fixed-route bus service is maintained, with the generalized travel cost of bus relative to cycling introduced as a passenger satisfaction constraint, and vehicle allocation and service frequency optimized to minimize operating costs. During off-peak periods, demand-responsive transit is introduced without increasing the system fleet size, and surplus capacity is reallo cated between fixed-route and demand-responsive services. A demand allocation coefficient is introduced to represent the passenger demand split between the two service modes, and a bi-objective optimization model is formulated to minimize operating service costs and passengers' generalized travel time costs by jointly optimizing the capacity allocation and service frequency of the two service modes. A single-objective genetic algorithm (GA) and a bi-objec tive non-dominated sorting genetic algorithm Ⅱ (NSGA-Ⅱ) are employed to solve the peak- and off-peak planning stages, respectively. An empirical study is conducted for the residential area surrounding Longchi Station of the Nanjing rail transit system. The results show that, during off-peak periods, compared with the pure fixed-route service mode, the benefit-optimal scheme of the integrated fixed-route and demand-responsive service mode increases vehicle utilization by 9.7%, while increasing per-capita operating cost by 2.13 CNY and reducing per-capita travel cost by 2.73 CNY. When the off-peak-to-peak passenger demand ratio α is no greater than 0.11 and the proportion β of passengers choosing fixed-route feeder services is less than 0.7, the integrated service mode achieves better overall benefits, demonstrating the effectiveness of the proposed model in improving resource utilization efficiency and service level of rail transit feeder bus systems.
Optimization Method of Guide Signs for Ten-lane Freeway Off-ramps based on Driving Task Analysis
LUO Qiang, MO Junhao, GUO Xiangyan, ZHAO Ding, CHEN Xinqiang
2026, 44(4): 138-149. doi: 10.3963/j.issn.1674-4861.2026.04.013
Abstract(25) HTML (12) PDF(0)
Abstract:
At exits of bidirectional 10-lane expressways, the increasing demand for continuous lane changes, coupled with insufficient adaptation between guide sign placement and driving task rhythm, tends to cause concentrated driving workload. This study proposes a guide sign optimization method based on driving task analysis. The continuous driving process from the mainline to the off-ramp is divided into five functional zones: mainline driving, first diversion, second diversion, deceleration adjustment, and ramp speed regulation. Five subtasks (visual scanning, lane control, sign identification, lane changing, and speed control) are decomposed and their workloads are quantified. Workload analysis reveals that the first and second diversion zones, with scores of 40 and 43 respectively, are the primary workload peaks during the exit process. Based on these findings, a "task workload-time-distance" mapping relationship is established to translate driving workload into spatial requirements for sign placement, enabling a transition from driving behavior analysis to the derivation of sign layout parameters. Accordingly, a six-level guide sign layout scheme featuring "cognitive advance and operational peak staggering" is proposed. A Unity driving simulation scenario was developed based on a real highway reconstruction and expansion project. Thirty-two drivers completed 160 valid trials across five guide sign schemes. By integrating drivers' information workload and vehicle control workload, and applying the entropy weight method for weight determination with subsequent scale calibration, a comprehensive Driver Workload (DWL) evaluation model was constructed. The results show that optimized Scheme E achieved a DWL of 0.331 ± 0.096, representing a 73.9% reduction compared to the actual engineering Scheme A (1.267±0.182), and was significantly lower than all other comparison schemes (p < 0.05), indicating that Scheme E effectively reduces overall driver workload and improves vehicle control stability. These findings demonstrate that zonal analysis based on driving task workload characteristics and optimization of sign spatial layout can enhance the matching between guide information and driving task rhythm, providing methodological support for the quantitative optimization of guide signs at exits of extra-wide expressways.
A Method for Measuring Bicycle-Sharing Tidal Flow Patterns from a Multi-level Perspective
CHEN Wendong, KANG Jiemin, LI Gen, WU Lan, CHENG Long
2026, 44(4): 150-160. doi: 10.3963/j.issn.1674-4861.2026.04.014
Abstract(27) HTML (10) PDF(1)
Abstract:
To address the modifiable areal unit problem (MAUP) caused by traditional aggregation units such as grids in shared bike tidal pattern detection, as well as the difficulty in balancing detection accuracy and computational efficiency in large-scale road networks, this paper proposes a two-stage tidal pattern measurement method from a multi-level perspective of "management subzone-road segment granularity." In the first stage, road segments are used as the basic detection unit instead of grids to avoid MAUP interference caused by subjective aggregation boundary delineation. In the second stage, the Leiden community detection algorithm is introduced to divide the study area into several management subzones with strong internal connections, shrinking the detection range of road-segment-scale tidal patterns from the entire region to within each subzone and achieving an order-of-magni tude reduction in computational complexity. Taking the shared bike system on Xiamen Island as a case study, the tidal patterns during the period from 06:00 to 10:00 are detected and analyzed. The results show that: in terms of accuracy, compared with grid-based methods, the two-stage method achieves significantly better accuracy, with the proportions of "extremely high" and "relatively high" accuracy reaching 74% and 61%, respectively, while the grid-based methods achieve only approximately 20% and 12%; in terms of computational efficiency, the two-stage method improves efficiency by 66% compared with direct processing of the entire region. A total of 83% of the tidal patterns occur during the morning peak (07:00—09:00), concentrated near employment and residential areas, respectively, exhibiting a clear unidirectional tidal flow pattern. The detected tidal patterns exhibit arbitrary shapes and multi-scale distributions that are difficult to capture with regular grid units, and tidal intensity varies significantly across time periods, indicating the necessity of graded dispatch strategies. In conclusion, the proposed method effectively overcomes the MAUP effect while balancing detection efficiency and identification accuracy, and can effectively enhance operational feasibility.
A Hybrid STGCN-Transformer-Bootstrap Method for Short-Term Expressway Traffic Volume Interval Forecasting
WANG Hefei, TENG Jing, YE Liang, AN Jian, CHEN Yuyi
2026, 44(4): 161-174. doi: 10.3963/j.issn.1674-4861.2026.04.015
Abstract(28) HTML (14) PDF(1)
Abstract:
Short-term traffic volume forecasting is a critical task supporting active traffic management strategies such as dynamic speed limits and ramp metering. To address the limitations of existing deep-learning models in modeling directional spatial propagation among cross-sections, suppressing noise during atypical operating periods, and quantifying point-prediction uncertainty, this study proposes an interval prediction method based on the STGCN-Transformer-Bootstrap model. A dual-stream parallel STGCN architecture is designed to decouple multi-horizon spatio-temporal features, and a gated cross-attention mechanism is introduced in the Transformer decoder to adaptively suppress the interference of irrelevant historical features. The Bootstrap method with 5 000 resampling times and the percentile method are then integrated to construct confidence intervals for the 5, 15 and 30 min prediction horizons. The method is validated on ETC gantry flow data from 26 cross-sections of the G15 Shenhai Expressway Quanzhou-Xiamen segment from June 1 to 15, 2022. At two typical cross-sections, compared with ARIMA and SVM, the RMSE, MAE, and MAPE of the proposed model decrease by 49.05%, 50.57% and 51.11% on average, with reductions ranging from 26.88% to 69.05%; compared with DCRNN, STGCN-LSTM, and GCN-Transformer, the three metrics decrease by 19.77%, 21.13% and 24.70% on average, with reductions ranging from 6.45% to 54.37%. Under the 95% confidence level, the percentile method achieves the best performance across the PICP, MPIW, and CWC metrics. Taking the 5 min prediction horizon as an example, MPIW is 59.24 pcu 5 min during stable free-flow periods, rises to 59.43 pcu/5 min during traffic phase transitions, and reaches 59.55 pcu/5 min during stable congestion, demonstrating adaptive interval widths that effectively characterize the uncertainty range under severe traffic flow fluctuations on expressways.

Copyright Editorial Office of Transport Information and Safety鄂ICP备05003336号-2

Address：No. 1178，Heping Avenue, Wuchang, Wuhan, CHINA (430063)

Tel：027-86580355 Email：jtjsj@vip.163.com

Supported by: Beijing Renhe Information Technology Co. Ltd