Your Destination for Mobility Engineering Resources

To regulate low-altitude UAV operations, this paper proposes a method for constructing a low-altitude layered route network. First, a visibility-graph-based obstacle environment modeling method is adopted to transform geometric constraints in complex low-altitude space into a graph representation, and candidate flyable networks are generated at different altitude layers. On this basis, a set of primary paths and low-overlap alternative paths is generated for multiple origin-destination pairs. To address the redundancy and discreteness of the resulting path set, multiple OD paths are combined into an initial low-altitude airspace route network through path screening, spatial aggregation, and shared-structure extraction. Furthermore, a minimum spanning tree is used to extract the network backbone for topological simplification. On this basis, redundant connections and reinforced critical route segments are introduced through network augmentation and optimization strategies, thereby constructing a low-altitude airspace network that integrates topological connectivity, structural compactness, operational efficiency, and redundancy robustness. Using Longhua District of Shenzhen as the case study area, this paper carries out simulation experiments on airspace planning. The results indicate that the proposed method requires only a marginal expansion of network size, yet it can markedly lower the average detour level of key OD pairs, enhance the two-path reachability ratio, and preserve stronger OD service capability under both random and directional disturbances. Therefore, the method is able to support the efficient construction of layered route networks in low-altitude environments characterized by dense obstacles, while also providing a methodological basis for the operational management of low-altitude UAV systems.
Liu, Mingyu, Zhang, Ying, Yin, Jianan, Tian, Wen, Meng, Fanyu
Against the backdrop of rapid urbanization, conventional bus networks with fully fixed stations often experience spatial mismatches between station locations and residents’ evolving travel demands, thereby reducing their service effectiveness. Demand Responsive Transit (DRT) can alleviate this mismatch by organizing services through a fixed and flexible stations scheme, in which some stations are retained as fixed anchors, whereas others can be served dynamically on demand. This study investigated fixed station selection for DRT using the existing station set of a conventional route as a candidate pool. We developed an evaluation framework that jointly considers time-dependent passenger flow characteristics and attractions induced by surrounding land use functions. Based on the traditional station impact index, we proposed an improved impact-index model by introducing a land use category weight coefficient to explicitly differentiate land uses with heterogeneous trip-generation potential. The resulting index integrates station-level ridership indicators, such as boardings, alightings and potential served passengers within a given time window, with land-use weights, enabling station ranking, grading, and threshold-based classification for fixed station retention. A case study of Harbin Bus Route 235 was conducted. Using weekday peak-hour smart card data, we estimated the station service demand and obtained passenger flow inputs for the proposed index. Land use information around each station was identified using Point of Interest (POI) queries and mapped to the corresponding weights. By computing and sorting the improved impact index and additionally checking the spatial distribution and inter-station spacing to preserve coverage and leave room for dynamic stations, a set of priority fixed stations for the peak period was determined. The results demonstrate the feasibility and effectiveness of the proposed method, providing an operable reference for scientific fixed-station planning and decision-making in DRT.
Feng, Yuxuan, Zhang, Xiqiao, Lin, Jiadong, Li, Yetong, Lu, Di
Accurately modeling intricate geographical interconnections and dynamic temporal patterns is still a significant difficulty in traffic forecasting, playing a crucial role in intelligent transportation systems. The majority of contemporary spatiotemporal graph neural networks are limited in their capacity to capture long-range temporal dependencies and time-varying spatial correlations because they rely on pre-defined or self-learning graph topologies in conjunction with recurrent-based temporal modeling. This study suggests a Time-Aware Spatio-Temporal Graph Neural Network (TASTGNN) for traffic forecasting in order to overcome these drawbacks. In order to enable the learned graph structures to dynamically adjust to changing traffic correlations over time, we first provide a time-aware graph learning technique that explicitly includes temporal context into spatial dependency modeling. Second, by using self-attention processes along both spatial and temporal dimensions, we develop an attention-enhanced spatio-temporal modeling framework that simultaneously captures long-range temporal dependencies and dynamic interactions among road network nodes. The efficacy of the suggested TASTGNN is shown by extensive tests carried out on two sample real-world traffic benchmark datasets.
Yin, Xueyan, Qian, Youcheng, Zhou, Wenze, Xu, Wenjie
To address the issue of weak constraints in the transverse and normal directions during space-based orbit determination for cislunar spacecraft, this paper proposes a space-based differenced measurement method. In this method, two Low Earth Orbit (LEO) satellites receive the same measurement signal beam transmitted by the cislunar spacecraft. By precisely measuring the time difference of signal arrival, angular constraints on the cislunar spacecraft are formed, thereby improving the convergence speed and accuracy in the transverse and normal directions of the orbit. The LEO satellites achieve high-precision baselines and relative clock offsets via common-view Global Navigation Satellite System (GNSS) signals. This method requires accurate calibration of the system delay between the two LEO satellites when receiving signals from the cislunar spacecraft. The differenced system delay is calibrated on-orbit using ground stations, achieving an accuracy of 593.7 ps. Taking cislunar spacecraft in both a Distant Retrograde Orbit (DRO) and a DRO direct transfer orbit as examples, this paper analyzes the impact of different LEO baseline lengths (500 km, 5000 km, and 10,000 km) and measurement modes (single-satellite, dual-satellite non-differenced, space-based differenced, and space-ground differenced) on orbit determination performance. Results from a dynamic batch least squares estimator show that for spacecraft in both the DRO and the DRO direct transfer orbit, single-satellite measurements fail to achieve rapid convergence of position estimation within 30 minutes. Under a 500 km baseline, the position estimation for the DRO using space-based differenced measurement converges to 200.3 m within 30 minutes, outperforming the dual-satellite non-differenced measurement (478.3 m). Under the same baseline condition, the position estimation for the DRO direct transfer orbit using dual-satellite non-differenced measurement converges to 2836.0 m, while the space-based differenced measurement achieves a significantly improved accuracy of 1013.0 m. Under a 5000 km baseline, the position estimation for the DRO using space-based differenced measurement converges to 30.1 m, outperforming the dual-satellite non-differenced measurement (84.6 m). Under the same baseline condition, the position estimation for the DRO direct transfer orbit using dual-satellite non-differenced measurement converges to 438.8 m, while the space-based differenced measurement achieves a significantly improved accuracy of 126.3 m. The space-ground differenced measurement fails to achieve convergence for position estimation of the DRO direct transfer orbit but converges to 208.5 m for the DRO. The space-based differenced measurement method can significantly improve the orbit determination convergence speed for cislunar spacecraft in either the DRO or the DRO direct transfer orbit. However, if the measurement arc exceeds one hour, the accuracy advantage of space-based differenced measurement is not significant.
Lai, Weifeng, Wang, Wenbin, Mao, Xinyuan, Feng, Xuan
This paper presents an integrated Attitude Determination and Control System framework for a low lunar frozen orbit mission, combining high-fidelity disturbance modeling with robust control design. The study addresses the challenges of the lunar environment, including external perturbations and limited reference availability, which impose strict requirements on estimation accuracy and control robustness. A realistic ADCS model is developed, accounting for solar radiation pressure, magnetic torque, gravity gradient effects and orbital conditions. Based on this model, estimation and control strategies are implemented to ensure reliable attitude acquisition and orbit maintenance. The ADCS performance is evaluated through a MATLAB® and Simulink® simulation framework under representative operational scenarios, including detumbling, slew maneuvers and Moon-pointing phases. In addition, a statistical robustness analysis is performed under different initial angular velocity conditions in order to assess the stability and adaptability of the proposed control architecture. Multiple simulation cases are analyzed by varying the initial angular rates and operational configurations, allowing the evaluation of convergence time, pointing accuracy, angular velocity decay and controller stability. Furthermore, the influence of eclipse phases on the ADCS behavior is investigated, highlighting the capability of the control system to maintain stable pointing performance despite temporary variations in environmental disturbances and sensor reference conditions. The results demonstrate that the proposed approach achieves stable and accurate attitude control despite environmental disturbances. In particular, the findings confirm the feasibility of effectively controlling a spacecraft operating in a low lunar frozen orbit, validating the robustness and reliability of the integrated ADCS design.
Di Domizio, Simone, Bernelli-Zazzera, Franco, Liang, Yuying
To enhance the robustness of satellite power systems and ensure overall spacecraft energy safety, this paper proposes a design scheme for a distributed high-power power supply control subsystem incorporating a Sequential Switching Shunt Maximum Power Regulator and a grid-connected controller. This configuration enables emergency power supply and peak power support under fault conditions, thereby improving solar energy utilization and load power reliability. To achieve intelligent grid connection among units, the battery voltage must be converted to a high voltage level with software-adjustable output voltage. For the hardware design of the grid-connected controller, a dual interleaved parallel Buck/Boost circuit plus a DC transformer is adopted as the topology of the bidirectional DC/DC converter, and the state-space averaging method is used to establish the small-signal model of the Buck/Boost circuit. A composite current sharing control strategy based on an analog and digital combined control with the main error amplifier and the droop current sharing method is proposed, which enables bidirectional power regulation through mode selection, avoiding output current polarization risks caused by differences in compartmentalization among power subsystems that lead to inconsistent output characteristics of individual grid-connected controllers. Meanwhile, in Boost mode, an improved algorithm is proposed to address the stability issues caused by conventional droop control. The droop coefficient is adjusted by compensating the bus voltage deviation, ensuring the system operates normally under heavy-load conditions. The composite control algorithm is simulated in MATLAB. Simulation results show that the proposed strategy enables flexible boost-type grid connection among subsystems and effectively improves the power supply reliability of the system.
Xu, Yiyuan, Xiong, You, Cao, Chengrong, Wang, Yongkang, Wu, Ziyue
Civil aviation air-to-ground (ATG) wireless broadband communication technology has become increasingly mature. Civil Aviation Flight University of China has long been committed to the R&D and application of this technology in general aviation aircraft, accumulating rich experience in aircraft modification and certification. This paper mainly introduces and analyzes the cases and designs of installing ATG systems on two types of aircraft: CE525 and domestic Xinzhou 600. It proposes that domestic avionics equipment development should meet the requirements of easy aircraft modification and system miniaturization, and presents the technical path for the miniaturization of airborne equipment, providing valuable references for domestic manufacturers in developing avionics products.
Xu, Xiaodong, Hou, Xiangyu, Gao, Jianwei, Kong, Xiangjun, Zhou, Feng, Chen, Qiang, Liu, Xiaoxi
The optimal design of speed limit signs is fundamental for ensuring driver comprehension and regulatory compliance on multi-lane freeways. Currently, two prevalent sign types—the Vehicle-Type-Based (VTB) scheme and the more information-intensive Lane-and-Vehicle-Type-Combined (LVTC) scheme—coexist in practice, yet a comparative understanding of their impacts on driver cognition remains lacking. To address this gap, a controlled indoor experiment was conducted from a human factors perspective. Fifty-two licensed drivers performed five types of simulated driving tasks (Simple/Difficult Lane Selection and Compliance Assessment, and Speed Compliance Judgment) under both passenger car and large truck scenarios. Driver performance was measured by response accuracy and reaction time. The results revealed a significant task-type dependency in visual recognition performance. For cognitively demanding decision-making tasks, the simpler VTB signs yielded significantly higher accuracy and shorter response times. Conversely, LVTC signs showed a limited accuracy advantage only in specific rule-verification tasks, but often at the cost of longer processing times. This pattern demonstrates a critical interaction mechanism between sign information complexity and task cognitive demand. Overall, the VTB scheme’s simpler structure supported more efficient cognitive processing across most situations. These findings provide an empirical, cognition-based rationale for prioritizing the VTB design on multi-lane freeways to reduce driver cognitive load and enhance traffic safety.
Di, Shengde, Wang, Deyu, Tang, Chengcheng, Wu, Hongbo, Guo, Da
In order to improve the state monitoring, fault reconstruction, and fault-tolerant control capabilities of the distributed control system (DCS) for aircraft engines, a fuzzy T-S model of the aero-engine DCS is established. Combined with sliding mode variable structure control theory, a generalized adaptive sliding mode observer(SMO) is designed to reconstruct fault signals from sensors and actuators. Secondly, based on the fault signal reconstruction, an active sliding mode tolerant controller is designed to the engine fuzzy T-S model under fault conditions. Finally, simulation results show that the system can readjust to a stable state within a short time under fault conditions, with a state deviation of less than 0.1%. The system can quickly readjust to a stable state, with a settling time of less than 1 second.
Zhang, Ledi, Fan, Wenlan, Liu, Zhixing, Zhao, Guixi, Jia, Lintong, Zhang, Shunde, Li, Peiyuan, Yang, Chengyin
Based on standardized C-NCAP active emergency braking (AEB) test data, this paper proposes an integrated performance evaluation method, termed the AEB Performance Score (APS), for time-series test data, aimed at quantifying the active safety performance of AEB systems under different target object types and initial speed conditions. The proposed method integrates key indicators, including avoidance outcome, TTC at AEB activation, maximum deceleration, speed reduction ratio, and stopping margin, and achieves unified evaluation of successful avoidance and collision-failure samples through normalization and hierarchical weighted mapping. Analysis based on real vehicle test data demonstrates that APS can effectively characterize AEB performance boundaries and differences across test scenarios. Correlation analysis further shows that APS is more strongly driven by the speed reduction ratio and stopping margin, indicating that early triggering alone is insufficient to ensure effective collision avoidance. The key to improving AEB performance lies in more accurate deceleration planning and execution consistency. This study provides a new analytical perspective for fine-grained performance evaluation and failure mode identification of AEB systems under standardized test conditions.
Li, Yupeng, Li, Kai, Zhang, Zhiqiang, Geng, Dongliang, Zhang, Shuai, Ma, Wenbo
The proliferation of four-wheel independent steering and driving (4WIS-4WID) electric vehicles (EVs) has introduced significant challenges for high-precision trajectory tracking control, including multi-actuator constraint handling, model uncertainties, and time-varying road conditions. To address these issues, this paper proposes a novel robust hierarchical control framework integrating Model Predictive Control and Quadratic Programming (MPC-QP) for real-time torque optimization in electric vehicle (EV) trajectory tracking. The upper layer employs a Tube-based robust MPC for high-precision path following and stability control, which explicitly handles bounded disturbances and model uncertainties via online road adhesion adaptation and robust invariant set design. The lower layer utilizes a sparse QP solver for optimal torque distribution among four in-wheel motors, with the objective of minimizing the tire load rate while strictly adhering to actuator and tire friction constraints. High-fidelity simulation results under double-lane change and steady-state cornering scenarios demonstrate that the proposed method reduces lateral tracking error by 45%, heading error by 38%, and overall energy consumption by 18% compared to a conventional Linear Quadratic Regulator (LQR) baseline. Crucially, the QP solver achieves a computation time of only 8 ms on an automotive-grade embedded platform, validating its real-time feasibility for Level 3 (L3) autonomous driving applications. The core innovations of this work lie in the decoupled design that enhances scalability, the online adaptation to uncertainties via Tube-MPC, and the efficient sparse formulation for real-time torque allocation, collectively advancing the robustness and efficiency of intelligent EV chassis control.
Zhu, Julian, Liu, Song, Zeng, Xin, Yang, Shichuan, Wang, Yongyuan, Liang, Dezhi
The current evaluation frameworks of feeder buses for metro have failed to adequately meet the diverse needs of passengers, deviating from the modern, human-centered transportation model that prioritizes user-centered design and service optimization. Our research seeks to develop an index evaluation model of feeder buses for metro, facilitating the distribution of optimization strategies from the viewpoints of both businesses and passengers. Our model builds on the traditional public transportation network evaluation index system by adding metro feeder indices with four levels of criteria: economic benefits, operational efficiency, service quality, and feeder capacity. There are also 16 evaluation indices. We did mining analysis on big spatiotemporal data, such as smart card data, schedules, GPS trajectories of vehicles, and the layouts of the routes and stations for feeder buses. We use data from Ganyintang Station on Metro #3 and its feeder buses to test how well our new model works. The results show that, based on our evaluation criteria, the evaluation scores of feeder buses are likely to go up by 26% if we improve the stations layout and scheduling of the feeder buses. Our research can help improve public transportation in big cities and make feeder buses work better.
Gong, Jiekun, He, Li, Deng, Jie, Xu, Tingxia
The Active-Imaging Satellite Mission Planning Problem (AI-SMPP) requires planners to manage task selection, scheduling, and observation durations simultaneously, while also balancing changing image quality against fluctuating energy limits. The problem builds on the classic Vehicle Routing Problem with Time Windows (VRPTW), but with a critical difference: AI-SMPP treats observation durations and image quality as variables rather than fixed values. This difference creates a tightly coupled, high-dimensional search space.We introduce a two-phase hybrid architecture called DPA-MHA (Dynamic Programming Approach integrated with Meta-Heuristic Algorithms) to handle this NP hard problem. Phase one fixes the observation parameters, reducing AI-SMPP to a standard VRPTW. A dynamic programming solver then establishes an optimal baseline sequence. Phase two relaxes those temporal constraints. At this stage, we use five meta-heuristics. Adaptive Large Neighborhood Search (ALNS), Ejection Chain, GRASP, Differential Evolution and Simulated Annealing to traverse the search space broadly. Each candidate generated by these heuristics then undergoes an exact refinement step via dynamic programming. Benchmark evaluations confirm a clear advantage for DPA-MHA. The proposed hybrid framework consistently surpasses standalone exact or conventional heuristic methods in computational efficiency, solution quality, and population diversity. Notably, the adaptive search module integrated with dynamic routing initialization demonstrates exceptional robustness in circumventing premature convergence to suboptimal solutions. Merging the accuracy of exact methods with the capacity to adjust to heuristics, this offers a viable strategy for high-dimensional aerospace operations. In such settings, standalone heuristics lack precision.
Chang, Kexin, Zhu, Zeqi, Ma, Runyu, Wen, Qi, Tang, Shiyang, Zou, Cong
Urban public facilities tunnels are vital underground infrastructures that transport electricity, communications, gas, water, and drainage systems. The safe and stable operation of these tunnels is directly related to the normal operation of the city and public safety. During long-term operation, these tunnels often face some common industry problems, such as the coupling of multiple risk factors, high uncertainty of accident occurrence, significant changes in dynamic risks, and lack of a large number of historical fault records. In order to solve these problems, this paper proposes a quantitative operation safety assessment method for urban public tunnels based on the fuzzy Bayesian network framework. By means of systematic literature review, field engineering investigation and semi-structured expert interviews, this method comprehensively identifies safety risk factors from four dimensions: structural integrity, equipment operation status, external environmental impact, and operation and maintenance management capabilities. A multi-level index system is constructed. On this basis, fault tree analysis is used to clarify the causal propagation logic and accident evolution path between risk factors, and the fault tree model is directly mapped to the Bayesian network topology. At the same time, combined with fuzzy set theory and expert experience, the expert evaluation in the form of language is transformed into trapezoidal fuzzy numbers. After defuzzification, the prior probability and conditional probability parameters of Bayesian network are obtained, which effectively solves the technical bottleneck problem of accurately obtaining parameters in the case of insufficient historical data. This method has been verified in a real urban public tunnel project. The overall safety risk probability is calculated by forward reasoning, and the key risk factors leading to the accident are located by reverse diagnosis technology. On this basis, targeted risk control measures are proposed. The research results show that this method can scientifically quantify the safety risk level of public tunnels, effectively solve the ambiguity and uncertainty problems in the evaluation process, and provide theoretical support and engineering reference for the safety management and risk control of urban public tunnels.
Zhang, Xinqiang, Xu, Teng, Zhu, Mingkun, Huang, Jiadi, Ba, Yunbing, Jiao, Guangxing
Lowering the rotational speed of a fan rotor helps reduce both fan noise and the structural strength requirements for the rotor. To achieve a reduction in the fan rotor's rotational speed, an ultra-highly loaded aerodynamic design is applied to an electric drive ducted fan rotor, utilizing a diffuser blade profile characterized by a high camber angle accompanied by low loss. This blade shape is employed to enhance the prototype fan rotor, enabling ultra-highly loaded operation at a low rotational speed. The findings indicate that with decreasing rotational speed, the load coefficient gradually rises, there is an increase in the rotor camber angle, and a decrease in the outlet relative flow angle. However, at the same time, the stator inlet Mach number increases, and the flow turning angle becomes larger, leading to greater flow losses and increased difficulty in stator design. While maintaining the flow rate and pressure ratio identical to the prototype rotor at the design point, the rotational speed is reduced from 3900 rpm to 1676 rpm (reduced to 43.0% of the prototype rotor’s rotating speed); at the design point, the rotor efficiency is 0.9450, and the stall margin is 127.76%, so it has high efficiency and a large stall margin. A stator matched with the rotor is designed to compose the fan stage with ultra-high load and low rotating speed. At the design point, the fan stage efficiency is 0.8766, with a total pressure ratio of 1.0201; the design targets are met, and the stall margin is very high at 53.73%. Due to the convergence of the rear part of the ducted fan rotor blade passage with ultra-highly loaded conditions, the flow separation is not easy to occur under high back pressure, and the stator’s angle of attack is small. Therefore, the stator separation on the suction surface is suppressed, and the stall margin is much higher.
Liu, Enbo, Zhou, Zhenggui
Tourist bus driver distraction and fatigue are critical driving risk factors, yet empirical studies are scarce due to data limitations. Based on approximately 7 million kilometers of DMS alert data from 692 tourist coaches, this study takes sub-trips as the analysis unit, develops a Gaussian Process Boosting (GPBoost) model with vehicle-enterprise grouped random effects to predict distraction/fatigue risks, and uses SHapley Additive exPlanations (SHAP) to interpret nonlinear variable effects. Results reveal predictive associations of continuous driving time and speed on both risks, interactive associations between speed and driving duration-risk relationships, and context-dependent direction-reversing effects of enterprise management factors across risk types. Vehicle-enterprise heterogeneities in grouped random effects are pronounced, calling for tiered risk management. This study provides a quantitative basis for proactive safety management of tourist buses.
Wu, Yaqi, Liu, Xiaobing
School escort travel is highly prevalent during the compulsory education stage and has emerged as a key contributor to traffic congestion in school-adjacent areas. To effectively guide and manage such travel behavior, it is essential not only to understand the factors influencing mode choice, but also to consider the often-overlooked sequential logic and heterogeneity in household decision-making processes. This study focuses on households of primary and secondary school students in Shanghai, and integrates multidimensional variables, including student attributes, parental employment status, household vehicle resources, and the built environment. Potential decision paths are identified through a correlation matrix, and differentiated escort path mechanisms are analyzed using structural equation modeling (AMOS) for working and non-working guardian groups. The results show that: (1) For working guardians, commuting constraints significantly weaken the impact of home-school distance on mode choice (path coefficient: 0.21 vs. 0.37 for non-working guardians escorting to school, and 0.28 vs.0.34 in the afternoon); (2) Among non-working guardians, the transmission effect along the path “household vehicle ownership → driving capability → travel mode” is stronger (path coefficients: 0.42–0.35 vs. 0.17–0.15 in working households in the morning; 0.40–0.37 vs. 0.22–0.22 for escorting from school). This study uncovers the dynamic, group-based decision-making mechanisms under spatiotemporal constraints and offers valuable insights for the targeted management and policy design of school escort travel.
Wang, Yanli, Zhu, Shengze, Zhang, Zhiwei, Xiang, Zhengliang
To address the problem of detecting aircraft ground icing under static conditions, this study proposes a non-contact aircraft ground icing detection method based on polarization characteristics in visible light. First, based on the Fresnel equations and the microfacet polarization bidirectional reflectance distribution function (pBRDF) model, the polarization properties of light at the interface of dielectric media were theoretically analyzed. The relationship between the degree of polarization (DoP), the complex refractive index of materials, and the observation geometry was derived, and a polarization optical model was established. The complex refractive indices of five typical aircraft skin coatings were experimentally measured in the visible spectrum, and the corresponding full-angle transmission curves of the degree of polarization (DoP)and angle of polarization(AoP) were obtained. On this basis, an icing detection system was developed using the Brewster angle of ice (52.7°) as the observation angle. Polarization images under both strong and weak illumination conditions were captured, and the differences in degree of polarization and angle of polarization (ΔDOP and ΔAOP) were calculated as key parameters for icing detection. A K-means grayscale clustering segmentation algorithm was employed to identify icing regions. Furthermore, a hierarchical decision model for determining icing location and type was established according to the ΔDOP and ΔAOP characteristic values of aircraft skin and different ice types (glaze ice, rime ice, and mixed ice). The results demonstrate that the proposed method achieves an ice type recognition accuracy of 98.6%. Compared with improved polarization reflection symmetric decomposition detection, polarization detection based on the Otsu algorithm, polarization scattering detection, and polarization-based transparent object segmentation, the proposed method shows significant advantages in ice type recognition accuracy and is more suitable for aircraft ground icing detection tasks.
Zhuge, Jingchang, Tang, Rongnian, Xing, Zhiwei
Regarding the issue in the 2024 version of C-NCAP’s automatic emergency braking (AEB) test where the scene score rates are highly concentrated at the maximum score, and traditional continuous regression modeling is difficult to effectively depict system risks, this paper transforms the AEB performance assessment problem into a rare failure risk prediction problem, and constructs an AEB failure risk assessment method based on LightGBM and SHAP. Based on 464 test samples, an explanatory variable system is established around operating condition parameters, vehicle configuration parameters, and process temporal sequence parameters, and samples with scene score rates lower than 1 are defined as failure samples. Firstly, the original data undergoes outlier removal, missing value handling, categorical variable encoding, and continuous variable standardization; secondly, a robust LightGBM classification model is constructed to predict AEB deduction/ failure risk; finally, the SHAP method is used to conduct interpretability analysis of the model to identify the key factors affecting AEB system performance. The results show that the established robust model achieves good recognition effects on the test set, with an AUC of 0.871, PR-AUC of 0.527, and F1 of 0.667. The feature importance and SHAP analysis results show that the number of sensors, relative speed, alarm-time TTC, braking-time TTC, overlap rate, and maximum braking deceleration are the key variables affecting AEB failure risk. Among them, higher sensor redundancy and earlier alarm and braking trigger times can significantly reduce system failure risk, while high relative speed, complex collision geometric relationship, and braking intervention lag will significantly increase the failure probability. The proposed method can effectively identify AEB system performance risks under the condition of highly concentrated score rates and rare failure samples, providing a data-driven analysis tool for AEB calibration optimization, sensor configuration design, and active safety performance improvement.
Yuan, Yue, Li, Yanchen, Zhang, Changlu, Li, Zhengwei
Due to topographical constraints and limited space for evasive manoeuvres, accidents in mountain expressway tunnels can easily trigger long-distance congestion, queues spilling onto the open road, and even a chain reaction of serious secondary accidents; To address this challenge, this paper explores a low-cost, easily implementable active control scheme. Focusing on single-direction, dual-lane mountain expressway tunnel sections, it proposes a coordinated active control strategy comprising ‘graded traffic restriction at the entrance + in-tunnel speed guidance based on the PSO algorithm’. Utilising the SUMO micro-simulation platform, a control group with no intervention was established for comparison against experimental groups with traffic restriction levels of 30%, 50%, 70% flow restriction levels for comparative validation. Additionally, the PSO algorithm was employed to determine the optimal speed limit combinations during the accident period. Following a multi-objective trade-off analysis of safety indicators and system delays, the study demonstrates that a 50% flow restriction threshold combined with specific segmented speed limits (46.5 km/h upstream of the incident and 19.8 km/h in the core incident zone) constitutes the multi-objective weighted optimal solution that balances both tunnel safety and overall road network efficiency; This strategy successfully validated the ‘congestion displacement’ mechanism, reducing high-risk queuing within the confined space from 600 m under uncontrolled conditions to 128 m. Furthermore, as this low-cost solution requires no complex equipment, it provides a useful theoretical basis and reference for the formulation of emergency management plans for mountain expressway tunnels.
Li, Youhe, Wu, Gang, Huang, Chongjian
To investigate differences in driving behavior across different sections of urban underground interchanges and to comprehensively evaluate their safety, this study selected six existing urban underground interchanges in China as case studies and constructed driving simulation scenarios to compare driving behaviors in merging, diverging, and basic sections. From longitudinal and lateral dimensions, four indicators—speed, acceleration, steering wheel angle, and lateral displacement—were selected, and the sample entropy method was applied to quantify safety differences in driving behavior among different sections. On this basis, an overall safety evaluation model based on an entropy-weighted improved grey near-optimal comprehensive evaluation method was proposed, and safety levels were classified using the Rank Sum Ratio (RSR) method. The results indicate that driving behavior safety entropy in merging and diverging sections is significantly higher than that in basic sections, implying lower overall safety. Moreover, longitudinal driving behavior safety entropy is slightly higher in merging sections than in diverging sections, whereas lateral driving behavior safety entropy is slightly lower. The proposed comprehensive safety evaluation model effectively reflects the overall safety levels of urban underground interchanges, with relatively balanced safety grade distributions across different evaluation dimensions. Among the evaluated cases, the Jiaozhou Bay Tunnel Interchange in Qingdao exhibits the highest overall safety level, whereas the Wanshishan–Zhonggushan Tunnel Interchange in Xiamen shows the lowest safety level and is classified as relatively dangerous, indicating the need for enhanced safety measures during the operation and management stage.
Tang, Xie, Yang, Yiqun, Lin, Guojin, Gao, Yanyang, Zheng, Liwen
Aiming at the problems of insufficient personalized adaptation and partial technologies relying on high-cost sensors in current new energy commercial vehicle energy-saving technologies, this paper conducts research on an intelligent recognition method of driver behavior for energy-saving driving of new energy commercial vehicles. The purpose is to realize low-cost energy-saving assisted driving of new energy vehicles by developing driving behavior recognition. Firstly, vehicle driving data covering various driving states from drivers with different driving styles are collected, and core characteristic parameters of driving behavior are extracted. Then, a driver’s driving style recognition model integrating feature extraction, PCA dimensionality reduction, preliminary clustering classification and LSTM-SVM is proposed to effectively recognize three types of driving styles: aggressive, steady and conservative. Based on this, a closed-loop energy-saving technical scheme of “recognition-adaptation-guidance” is proposed, and differentiated energy-saving optimization strategies are designed for different driving styles. Joint simulation verification shows that the application of this energy-saving scheme can save about 9.1% of energy consumption. The research technology in this paper can carry out driver behavior recognition without relying on high-cost sensors, and is suitable for low-cost new energy logistics vehicles.
Zhang, Hao, Wu, Yutong, Yu, Hanzhengnan, Li, Jingyuan, Wang, Fengbin, Liu, Yu