Lane-changing risk significantly increases on icy and snowy surfaces, resulting in frequent accidents. However, current lane-changing risk prediction models fail to incorporate driving intention recognition, which limits their accuracy and practical applicability. Therefore, this study proposes a dynamic lane-changing risk prediction framework for icy and snowy surfaces that incorporates driving intention recognition. Driving simulation experiments were designed to replicate icy and snowy driving conditions, and the resulting simulation data were used for model training and testing. First, a Bidirectional Multi-layer Long Short-term Memory (Multi-BiLSTM) network was employed to recognize the drivers' lane-changing intentions. Second, lane-changing risk was analyzed from both temporal and spatial dimensions, quantified using fault tree analysis and the Lane-changing Risk Index, and categorized using a k-means clustering algorithm. Finally, the Light Gradient Boosting Machine (LightGBM) algorithm was applied to predict lane-changing risks. Results indicate that the average duration of lane-changing intentions on icy and snowy surfaces was 6.12 s, a 36.3% increase compared to normal road surfaces. The Multi-BiLSTM model achieved recognition accuracies of 98.22%, 97.74%, and 96.31% for left lane-changing, right lane-changing, and lane-keeping, respectively. The LightGBM model achieved an overall accuracy of 97.49% in predicting lane-changing risk, outperforming other machine learning algorithms. These findings provide theoretical support for developing risk warning systems and control strategies for intelligent vehicles on icy and snowy surfaces.

In low-friction emergency braking scenarios (e.g., snow/ice), uncertainties in the brake command chain—such as reaction delay and actuator dynamics—can cause substantial variability in stopping distance, making deterministic single-run evaluation insufficient for high-quantile risk characterization and graded conclusions. This paper proposes a reusable closed-loop framework of ”LHS sparse sampling–GP surrogate–Monte Carlo propagation–risk mapping.” A CarSim–Simulink co-simulation is used as the high-fidelity truth model. Under the Step→Transport Delay→Rate Limiter→Saturation pressure command chain, four interpretable uncertain inputs are considered: driver reaction time τ_r, brake build-up delay τ_b, pressure ramp time T_ramp, and maximum master-cylinder pressure P_max. A unified post-processing convention is adopted to extract the total stopping distance S_total and the braking-only distance S_brake. Large-scale uncertainty propagation is then performed on the surrogate to obtain distributions and key quantiles. A five-level grading rule is constructed using global quantile thresholds, and a five-level risk map is generated via a conservative conditional q_0.95 statistic to localize high-risk input regions. The surrogate achieves centimeter-level accuracy on an independent validation set, and the propagated results provide stable high-quantile levels of S_total and S_brake, offering interpretable quantitative evidence for parameter selection, test benchmarking, and risk grading under low-friction braking.

Reliable measurement of tire-road friction coefficient is essential for automated and active-safety functions on winter roads, where the available adhesion is low, spatially heterogeneous, and may change abruptly. This study proposes a real-time measurement-oriented fusion method that integrates front-view visual perception with vehicle lateral dynamics to quantify the tire-road friction coefficient on snow- and ice-covered roads. A SegFormer-B5 semantic segmentation network identifies road-surface classes from onboard images, and a road type-friction range mapping converts the class probability distribution into a physically interpretable visual prior with equivalent uncertainty. In parallel, a 3-DOF bicycle model and a brush tire model are combined with IMU acceleration, wheel-speed, yaw-rate, and steering-angle measurements. An EKF estimates longitudinal and lateral velocities, while a two-stage UKF sequentially estimates tire lateral forces and identifies the front- and rear-axle friction coefficients. To improve measurement reliability, spatiotemporal synchronization aligns the visual prior with the tire-road contact region, and an adaptive confidence-weighted fusion strategy incorporates segmentation entropy, slip angle, lateral acceleration, and posterior covariance. Consistency gating and friction-circle constraints are further imposed to reject abnormal estimates and maintain physical feasibility. CarSim-Simulink co-simulations under two segmented adhesion-transition scenarios show MAEs of 0.007432 and 0.000398 and RMSEs of 0.008711 and 0.001468, reducing MAE by 73.29% and 92.91% compared with dynamics-only estimation. These results demonstrate that the proposed method can provide accurate, robust, and physically consistent friction measurements for ABS, ESC, AEB, and automated driving decision-making under winter road conditions.

To prevent collisions with a preceding vehicle under winter road conditions in cold regions, this study investigates longitudinal braking collision avoidance, lateral lane-change collision avoidance, and a decision-making strategy for collision avoidance modes of autonomous vehicles. Using a pendulum friction tester, the road adhesion coefficient on winter roads in cold regions was measured. Taking this coefficient into account, longitudinal and lateral safety distance models were established, and a collision avoidance mode decision strategy was designed based on the time-to-collision (TTC) model and the safety distance model. For longitudinal braking collision avoidance control, a hierarchical control approach was employed to develop a longitudinal graded braking controller that considers the road adhesion coefficient. For lateral lane-change collision avoidance, an adaptive-horizon model predictive control (MPC) controller was designed, with the horizon varying according to the road adhesion coefficient and the ego vehicle speed. Finally, the proposed collision avoidance control system was simulated and evaluated on a co-simulation platform. In the longitudinal collision avoidance system, scenarios were established including a braking preceding vehicle and a sudden change in adhesion coefficient, under various road adhesion coefficients and vehicle speeds The distances to the preceding vehicle after avoidance were 1.8 m, 1.095 m, and 2.74 m, respectively, confirming the rationality of the graded braking collision avoidance design. When longitudinal braking is insufficient to prevent a collision and the inter-vehicle distance falls between the minimum braking distance and the critical lane-change distance, the system activates the lateral collision avoidance mode. On medium- and low-adhesion road surfaces, the MPC controller accurately tracks the lane-change trajectory to complete the avoidance maneuver while maintaining satisfactory driving stability.

Highway tunnels are locations where traffic accidents frequently occur; thus, accurately predicting and analyzing these incidents is crucial for alleviating traffic congestion and enhancing operational management. This article utilizes data from the Norwegian National Road Public Database spanning 2010 to 2020 to conduct a multifaceted feature analysis of traffic accidents in 274 tunnels, focusing on severity, spatiotemporal distribution, and vehicle type. Initially, ten highly relevant influencing factors were identified through a chi-square test, and the association rule method was employed to uncover intrinsic correlations among the variables. Subsequently, the SHAP (Shapley Additive exPlanations) method facilitated an interpretability analysis, elucidating the contributions and nonlinear impacts of factors such as weather, road conditions, and vehicle types on accident severity, while also revealing the interactive effects among key variables. For accident level prediction, a random forest model was developed, with a confusion matrix and ROC (Receiver Operating Characteristic) curve serving as evaluation metrics for comparison with Bayesian networks and XGBoost (eXtreme Gradient Boosting). The findings indicate that the random forest model outperforms the others, achieving an accuracy of 86

Traffic sign detection is widely used in automatic driving, assisted driving, and other intelligent transportation systems, and its detection performance is closely related to rain-driving safety. In the process of traffic sign detection, there are traffic signs of different sizes with large size variations, which will have a certain impact on the detection accuracy. For the problem that the existing target detection algorithm is poor in detecting traffic signs with small sizes, low resolution, and inconspicuous features in the image, a traffic sign detection algorithm based on improved YOLOv5s is proposed. The M-ASPP module is proposed to enhance the perceptual field without reducing the feature information and increase the feature information fusion at different scales to improve the traffic sign detection accuracy. Introduce BiFPN instead of FPN to enable the fusion of deep high-level semantic information and shallow location information to improve the recognition accuracy of the algorithm. Add a small target detection head to improve the detection accuracy of small target traffic signs. Use the NWD loss function insheathe d of the IoU loss function to reduce the small-target scale sensitivity problem and improve the similarity between small-target detection. The improved traffic sign detection algorithm shows better detection performance in snowy environments, taking into account both detection accuracy and detection speed.

The energy absorbing box plays the role of buffering energy absorption in the process of frontal collisions of automobiles, but the current structure of the energy absorbing box is single, and the energy absorption effect is limited. In this study, a honeycomb structure was filled into a thin-walled square tube to design a new type of automotive energy absorbing box. Firstly, a frontal collision finite element model is established with a vehicle model as the research object, and the crashworthiness defects of the vehicle model under frontal collision conditions are pointed out. Next, two filling forms of new energy absorbing boxes were designed, and the crashworthiness was compared through impact simulation, followed by multi-objective optimization to further enhance their performance. Finally, the original thin-walled square tube energy absorbing box was replaced with the honeycomb-filled design. A simplified model was used to compare the crashworthiness of the vehicle before and after the replacement. The results showed that, after the replacement, the vehicle's acceleration decreased by 11.49

In order to reduce the possibility of collisions during the driving process of intelligent vehicles in the same direction, this paper studies the collision avoidance control of intelligent vehicles in the same direction and designs an active collision avoidance controller. The longitudinal safe distance model, lateral lane change path planning model, and adaptive multi‐point preview model of preview distance are established. The longitudinal speed control is carried out by the expert PID control method based on mode switching, the lateral path tracking control is carried out by the sliding mode control method with exponential convergence law, and the active collision avoidance controller is designed in combination with the multi‐point preview module that is adaptive to the preview distance. The active collision avoidance controller was jointly simulated using Carsim, Prescan, and Simulink software for emergency lane change scenarios and slow vehicle driving in front. In the emergency lane change scenario, the minimum distance between the two vehicles is 1.9 m, and the path tracking deviation is 0.17 m. In the front vehicle slow driving scenario, the minimum distance between the two vehicles is 2.2 m, and the path tracking deviation is 0.13 m. The controller can realize collision avoidance in two scenarios of 80 and 108 km/h respectively, which shows that the controller is robust and considers the tracking accuracy and steering stability at the same time, which is of reference significance for improving the safety of intelligent vehicles driving in the same direction.

The impact resistance and energy absorption characteristics of the existing traditional honeycomb composite structures have not been fully explored, and in order to improve its mechanical performance under low-velocity impact loading, reasonable improvement is needed. In this study, an improved honeycomb composite structure is proposed by introducing a styrene-butadiene rubber (SBR) elastomer interlayer between the CFRP skin of the traditional composite structure and replacing the traditional hexagonal aluminum honeycomb core with a bio-inspired pomelo peel-like aluminum honeycomb core (Hg). First, drop-weight impact tests were conducted at 10 J impact energy to compare the impact resistance and energy absorption characteristics between traditional and improved honeycomb composite structures. Second, finite element models of the improved structure with five varying thickness configurations were established to analyze the influence of CFRP and SBR thickness on the impact resistance. Additionally, numerical simulations were performed by replacing the original hexagonal honeycomb core with Hg cores of varying aperture-to-wall thickness ratios (β) to analyze the influence of β on the energy absorption characteristics. The results demonstrated that the honeycomb composite structure exhibited significantly enhanced mechanical performance after the improvements and core replacement. Specifically, the impact resistance showed a strong dependence on the thickness of both the SBR interlayer and CFRP skin, while the energy absorption characteristics were closely related to the β of the Hg core.

Plain woven carbon fiber fabric offer significant potential for the lightweight design of automotive body parts; however, it also introduce challenges in the molding process, particularly in controlling molding defects. To address these challenges, this study investigates the molding defect characteristics of plain woven carbon fiber fabric through experimental and numerical simulations, proposing a multi-objective optimization method to improve fabric formability. First, the mechanical properties of plain woven carbon fiber fabric was evaluated using the picture frame shear test, and the key factors influencing it formability was identified. Second, using the B-pillar local reinforcement plate as a case study, the effects of blank holding force magnitude, as well as the length and width of the blank holding region, on the fabric's formability were examined. Additionally, the distribution patterns of the shear angle and fiber-directional strain under blank holding conditions were analyzed. Finally, a multi-objective optimization method based on the NSGA-II algorithm was developed to enhance fabric formability. The results demonstrate that as the shear angle increases, the wrinkling strain along the fiber direction rises significantly. Through optimization, the maximum axial angle and fiber strain were reduced by 85.5% and 99.3%, respectively, significantly minimizing molding defects in the B-pillar local reinforcement plate and improving the component's formability.

Honeycomb, as a thin-walled structure, has excellent energy absorption characteristics and is widely used in aerospace and automotive industries. In this paper, the quasi-static compression experiments of traditional hexagonal aluminum honeycomb structures with different apertures are carried out under axial load. The stress-strain curves and deformation modes are analyzed, the specific energy absorption of the structure is calculated, and the influence of the aperture on the energy absorption characteristics is determined. Based on the hybrid strategy, five new bionic honeycomb structures evolved from the traditional hexagonal honeycomb structure were designed, and a quasi-static compression simulation model was established for the new bionic honeycomb structure. The reliability of the simulation model was verified by the experimental results. The results showed that the energy absorption characteristics of the five new bionic honeycomb structures were superior to the traditional hexagonal honeycomb structures. Among them, Hg type bionic honeycomb structure has the best energy absorption characteristics, and its specific energy absorption is 16.93 J/g. Using the sensitivity analysis method, it was determined that the wall thickness t and the aperture ρ had the greatest influence on energy absorption. A mathematical optimization model was established, and NSGA-II algorithm was used to solve the Pareto optimal front solution to determine the optimal optimization scheme. The specific energy absorption of the optimized Hg bionic honeycomb structure was increased by 3.90

Traffic sign detection is important in intelligent transport systems such as autonomous and assisted driving. However, traffic sign detection suffers from a slight occlusion problem in snowy environments, which leads to long computation time of the detection algorithm and unsatisfactory detection rate. In order to solve these problems, this paper introduces the Ghost module to replace the Bottleneck module in the C3 module to obtain the C3Ghost module in the YOLOv5s model to reduce the computational redundancy and the number of parameters, and improve the inference speed. Secondly, the CA attention mechanism is introduced into the neural network to enhance the regression and localisation ability of the model by embedding the location information to extract important features, so as to improve the ability of the network to extract accurate location information. And the NWD loss function is used instead of the IoU loss function to improve the detection accuracy and stability of the model and ensure that the model can better capture and distinguish small target features. By comparing the results of the TT00k dataset with YOLOv5s, the computational loads (CLOPs) of the improved model are reduced by 22.5

Occupant restraint systems (ORS) play an integral and critical role in protecting occupant safety. However, research on optimising ORS in reclined postures in autonomous vehicles is still insufficient. This study aimed to conduct multiobjective optimisation of the ORS to explore potential occupant injury reduction capability. Firstly, the MADYMO simulation method quantitatively correlates the relevant parameters of the ORS with occupant injuries. Then, parameters that significantly impact the safety performance of the ORS (pretensioning time, pretensioning force, webbing stiffness, seatbelt limiter force, and explosive pretensioning time) were selected as design variables. The occupant's comprehensive injury evaluation criterion, WIC, was used as the optimisation criteria. The ORS parameters were optimised by combining the Latin hypercube experimental design, the Kriging surrogate model, and the NSGA-II optimisation algorithm. The results show that after optimisation, the occupant's head injury criteria (Chead) decreased by 10.87%, neck injury criteria (Cneck) decreased by 14.18%, chest injury criteria (Cchest) decreased by 8.43%, lumbar spine injury criteria (Clumbar) decreased by 2.37%, and the comprehensive weighted injury criterion (WIC) decreased by 8.14%. This study significantly reduced injuries to various parts of the occupant, thereby improving the safety of occupants in reclined postures.

Optical communication has become a research hotspot in modern non‐contact communication systems, and high‐performance photodetectors, as a key component, are crucial for achieving high‐fidelity signal transmission. Photoconductive devices (PCDs) are expected to find applications in the field of optical communication due to their advantages in high gain. However, it is difficult to achieve an effective balance between high responsiveness and fast response times in PCDs, leading to significant challenges in their commercial applications in the field of optical communication. In this paper, by employing both the mask method and channel width tuning method, a high‐performance PCD is developed based on the Ag‐MAPbI 3 ‐Ag structure, demonstrating a high responsivity of 45.5 A W −1 , a high detectivity of 7 × 10 12 Jones, and a fast response time of 0.35 ms. Finally, this device is integrated into an optical communication system using embedded technology, effectively achieves byte transmission, and demonstrates the practical application of single‐crystal MAPbI 3 PCD as a signal receiver in the system. This research provides a noteworthy approach to facilitating the seamless integration and commercialization of optical communication systems utilizing perovskite single‐crystal devices, while also paving the way for optical interconnects based on perovskite.

The joints in an automobile’s body structure are crucial in bearing loads and transmitting stresses, thereby significantly affecting the body’s rigidity. To effectively improve body rigidity and crashworthiness, this study employed a sensitivity analysis to identify the critical joints among the nine joints of a specific sport utility vehicle (SUV) body. Following regulatory requirements, collision simulations were performed, revealing that the joint below the B-pillar exhibited the most significant deformation. Thus, using the material and thickness of the B-pillar’s lower joint as design variables, experimental samples were generated by the design of experiment (DOE). A multi-objective optimization for the B-pillar’s lower joint model was conducted using the response surface method and the simulated annealing algorithm to determine the final optimized solution. The optimization results showed a 9.31% increase in body bending stiffness, an 11.37% increase in torsional stiffness, and reduced intrusion at various points on the B-pillar, effectively enhancing the body’s rigidity and crashworthiness.

Hybrid perovskite CH3NH3PbCl3 single crystals have received much attention in the field of photodetectors due to their outstanding optoelectronic properties. Herein, a simple growth method of CH3NH3PbCl3 single crystal sheets was demonstrated in detail. The steady-state solution micro-reaction method with the small-hole rate control and the liquid-surface height control effectively solves the disadvantages of liquid-and gas-phase methods and realizes the preparation of a CH3NH3PbCl3 single crystal sheet with the controllable growth rate, size, and thickness, and good crystal quality and high crystallinity. Moreover, photodetectors based on one CH3NH3PbCl3 single crystal perovskite were fabricated with the responsivity of 80 mA/W under 395 nm light at 5 V. Meanwhile, a rise time of 22.3 mu s and a fall time of 16.5 mu s were obtained, indicative of a relatively fast response. These findings demonstrate that the CH3NH3PbCl3 single crystal perovskite have great potential applications in optoelectronic fields. (c) 2023 Elsevier B.V. All rights reserved.

针对高速公路匝道和公路急转道等处交通事故频发,防护栏碰撞缓冲吸能不足的问题,设计了一种新型蜂窝结构旋转式公路防撞护栏,建立有限元模型,运用仿真分析软件LS-DYNA模拟1.5 t小型客车、10 t货车模型与护栏的碰撞过程.根据仿真结果中出现的护栏横向变形较大、旋转桶倾倒等问题,采用正交试验对旋转护栏结构重要参数进行优化,得到新型旋转护栏最优方案,并分析最优方案下护栏的防撞性能.结果表明:通过正交实验优化后的新型旋转护栏在小型客车、货车碰撞下,X、Y方向质心最大加速度均小于20 g,护栏动态变形小于1000 mm,碰撞过程中没有出现车辆翻越、骑跨、下穿护栏的现象,对车辆有着较好的阻挡、缓冲、导向功能,可有效保障乘员行驶安全.

为了分析散落物在汽车发生碰撞时的分布特征及不同材质类型的散落物对分布特征的影响,为汽车事故再现中的碰撞速度推理提供参考,设计了智能循迹货车和乘用车模型的实车碰撞试验.通过两种类型车辆的完全侧面碰撞试验,模拟分析了事故现场散落物分布面积、质量与碰撞速度之间的关系.研究结果表明:当发生侧面碰撞时,不同材质类型的散落物具有自身的散落特性,其中大豆的散落面积最大,瓜子的散落质量最多,同种材质散落物的散落面积、质量与碰撞速度有着显著的数学关系.

针对未来智能车辆上乘员坐姿多样化,自动紧急制动系统(AEB)的应用会对乘员造成碰撞前的离位现象,对乘员施加主动预紧,研究不同坐姿下安全带对乘员的保护性能.通过建立主动人体模型、自动紧急制动系统相结合的约束系统模型,构建乘员前倾、后倾4种坐姿,设定主动预紧器中预紧力和预紧时刻的参数范围,以乘员头、胸、背的离位距离为研究目标,建立Kriging代理模型,并采用NSGA-Ⅱ优化算法对主动预紧器参数进行优化.结果表明:各坐姿下通过优化算法得到的最优预紧力和预紧时刻,运用Madymo仿真计算离位距离结果与预测值的误差在7％以内,乘员在制动时的离位距离有所改善.OOP1和OOP2坐姿中离位距离与预紧力呈负相关,OOP3和OOP4坐姿与预紧时刻呈现正相关,前倾角度大的坐姿预紧力越大,后倾坐姿则反之,预紧时刻触发越早对4种乘员坐姿的约束越有利.

The COVID-19 outbreak-caused blockade and disruption of the supply chain have dramatically increased the prices of perishable food and other products that rely heavily on the timeliness of supply chains. In the case of inflation, this study aims to make some adjustment to the pricing and replenishment strategy of perishable food and compare it with the scenario without considering inflation to determine the impact of the inflation rate, quality deterioration, time value of money, and characteristics of cash flow of perishable food sales on the supply chain decision-making. We used the discounted cash flow (DCF) model to measure retailers' revenue, which established that the optimal pricing and replenishing strategy could maximize the retailers' profit. Besides, the findings were compared with the traditional profit model. Moreover, numerical experiments and sensitivity analysis were provided for decision support to retailers. Overall, this study validates that inflation significantly affects the pricing and replenishment strategy, and the DCF model is more suitable to evaluate the profits of perishable food.