Current Issue

• 2025, Vol. 16 No. 2
  Published on:30 April 2025 Previous issue   
  Review, Progress and Prospects

  Review on driving risk monitoring and intervention technologies

  LI Guofa, OUYANG Delin, CHEN Chen, NIE Binging, ZHANG Wei, YU Huili, Liu Bin, ZHANG Qiang, WANG Wenjun, CHENG Bo, LI Shengbo

  2025, 16(2):  181-196.  doi:10.3969/j.issn.1674-8484.2025.02.001

  Abstract ( 16 )   HTML ( 1)   PDF (1621KB) ( 10 )  

  Safety has always been a critical concern in road transportation, serving as a fundamental pillar for ensuring traffic efficiency and supporting economic development. Driving risk monitoring and intervention are key technologies for enhancing vehicle safety, particularly with advancements in perception and information technology, which provide a robust data foundation and new avenues for implementation. This paper systematically reviews the research progress of driving risk monitoring and intervention techniques. Firstly, it examines the current state of driving risk monitoring from both of in-vehicle and external perspectives. Secondly, it reviews intervention strategies from both offline and online approaches. Studies have shown that interventions integrating visual, auditory, and haptic feedback can significantly improve driver response times, while haptic warning systems can help reduce the rate of driver errors. Then it is explored that the integration of risk monitoring and intervention technologies into Advanced Driver Assistance Systems (ADAS), autonomous driving systems, connected vehicle systems, and automated driving platforms. Studies have shown that intelligent systems based on vehicle-road-cloud collaboration can improve the real-time performance of risk warnings. The application of ADAS has been proven effective in reducing traffic accident rates and lowering Usage-Based Insurance (UBI) loss ratios. Finally, future research directions are discussed, including model optimization for lightweight deployment, big data applications, cloud-based control platforms, and the role of large-scale autonomous driving models in advancing risk monitoring and intervention technologies.

  Figures and Tables | References | Related Articles | Metrics

  Automotive Safety

  Compensation method for tooth surface measurement error of spiral bevel gear of automotive drive axle

  LIU Yongsheng, TAN Jiamin, WANG Ruifu, HU Panru, GAN Xinbin, CHEN Yixin

  2025, 16(2):  197-206.  doi:10.3969/j.issn.1674-8484.2025.02.002

  Abstract ( 10 )   HTML ( 1)   PDF (2047KB) ( 4 )  

  In order to ensure the safety and energy saving performance of the vehicle, it is crucial to improve the machining quality of spiral bevel gear tooth surface. An Iterative Closest Point (ICP) error compensation method optimized by dual quaternion was proposed for the measurement error of the measured and theoretical tooth surface. The error compensation problem was transformed into the matching of two surfaces. Dual quaternions were used to represent the tooth surface matching model, which helped to obtain the error matrix. The error matrix was linearized and a convex relaxation global optimization algorithm was applied to optimize the real part of the matrix. And then the precision matching of the spiral bevel gear tooth surfaces was achieved. The results show that the error compensation for the concave tooth surface reaches up to 77%. Specifically, the maximum error is reduced from 22.11 μm to 5.64 μm and the average error is reduced from 10.34 μm to 2.38 μm. Compared with the traditional Singular Value Decomposition (SVD) method, Quaternion method and Levenberg-Marquardt (L-M) method, the proposed algorithm has higher accuracy and stability, proving that the proposed compensation method is feasible.

  Figures and Tables | References | Related Articles | Metrics

  Research on head-neck injury of far-side occupant in side pole impact of electric vehicle with dual front passengers

  LÜ Yuanpeng, WANG Fang, LONG Chunguang, WANG Danqi, ZOU Tiefang, LIU Yu

  2025, 16(2):  207-216.  doi:10.3969/j.issn.1674-8484.2025.02.003

  Abstract ( 10 )   HTML ( 2)   PDF (5970KB) ( 5 )  

  To investigate the impact of occupant size differences and mutual interactions on the far-side occupant in side pole collisions involving electric vehicles, this study used a 5th percentile female as the near-side occupant and a 50th percentile male as the far-side occupant, and constructed various simulation scenarios by altering the collision angle and position. A linear fitting method was employed to numerically analyze the kinematic responses and head and neck injuries of the far-side occupant under different collision conditions. The results show that with the collision angle increasing, the lateral displacement of the far-side occupant increases, the restraining effect of the seatbelt weakens, and the occupant is more likely to collide with the near-side occupant or themselves. When the collision angle exceeds 45°, the HIC₁₅ predicted AIS 3+ injury risk surpasses 50%. The Head Injury Criteria (HIP) values indicate that, in all cases, the head absorbs a significant amount of energy, suggesting a high risk of AIS 3+ traumatic brain injury for the far-side occupant. Neck anterior longitudinal ligament (ALL) injuries predominantly occur in high-angle collisions and are correlated with the collision angle. Additionally, the posterior longitudinal ligament (PLL), capsular ligament (CL), and interspinous ligament (ISL) show a significant risk of neck ligament injuries in almost all cases.

  Figures and Tables | References | Related Articles | Metrics

  Multi-objective torque distribution strategy for hub motor electric vehicles under emergency steering conditions

  FAN Xiaobin, PENG Jiaxing

  2025, 16(2):  217-225.  doi:10.3969/j.issn.1674-8484.2025.02.004

  Abstract ( 9 )   HTML ( 1)   PDF (2666KB) ( 4 )  

  A multi-objective optimization strategy based on the non-dominated sorting genetic algorithmⅡ (NSGA-Ⅱ) was developed to improve steering stability and occupant comfort in hub-motor electric vehicles under emergency steering conditions. The strategy adopted a direct yaw moment control method to design the optimized objective function for handling stability and occupant comfort, taking into account generalized force equation constraints such as longitudinal desired torque and additional yaw moment. The optimal torque solution was selected in the pareto front calculated by NSGA-Ⅱ according to the dynamic conditions of the vehicle. The results show that compared to the axle load ratio (ALR) and weighted least squares (WLS) strategies, for high-speed steering, the root-mean-square errors of the vehicle sideslip angle and the yaw rate of the optimization strategy with respect to the reference values are reduced by 37.45%, 52.08%, and 41.98%, 56.95%, respectively, and the amplitudes of its lateral acceleration and torque are also smaller. The effectiveness of the proposed strategy for torque distribution during emergency steering of hub-motor electric vehicles is demonstrated to improve the vehicle steering stability and occupant comfort.

  Figures and Tables | References | Related Articles | Metrics

  Driver fatigue detection based on functional brain networks and graph convolutional networks

  XU Junli

  2025, 16(2):  226-233.  doi:10.3969/j.issn.1674-8484.2025.02.005

  Abstract ( 10 )   HTML ( 1)   PDF (1553KB) ( 9 )  

  To address the issue of ambiguous threshold criteria in constructing functional brain networks (FBN) for fatigue detection, this paper proposed to set a fixed threshold and employing graph convolutional networks (GCN) to optimize the learning of brain network graph features. A threshold of 0.5 was set for building the FBN, and the degree and clustering coefficient features of the network were extracted. These features were then input into the GCN, which learned and optimized the graph features for detection classification. The results show that the preposed model's detection accuracy has reached 88.90%. Furthermore, degree centrality identifies 14 significant electrodes within the brain network. Among them, the GCN model built on 7 key electrodes achieves an 87.2% detection accuracy, with faster detection speed and superior overall performance compared to the detection model based on 30 leads.

  Figures and Tables | References | Related Articles | Metrics

  Automotive Energy Efficiency and Environment Protection

  Unsupervised learning early warning of lithium battery failure driven by cloud data

  ZHOU Zhengyi, YANG Lin, MENG Yizhen, LI Huaijin, LÜ Feng, LIU Zhisheng, LI Yang, WU Weikun

  2025, 16(2):  234-242.  doi:10.3969/j.issn.1674-8484.2025.02.006

  Abstract ( 7 )   HTML ( 2)   PDF (1463KB) ( 3 )  

  An unsupervised learning early warning method was proposed based on voltage consistency to warn early the lithium battery faults in cloud battery management technology. The voltage characteristics in the effective charging cycle were extracted with measuring the degree of voltage consistency by using a minimum neighborhood radius which achieved a single cluster number for DBSCAN (density-based spatial clustering of applications with noise); A parameter with dimension-one was defined to improve the algorithm generalization ability to the actual working conditions; The hyperparameters such as alarm thresholds were selected through orthogonal experiment. The actual fault cases were verified and analyzed. The results show that for the battery systems with the low state of charge (SOC) faults, the single battery undervoltage faults, and the single consistency faults, this method enables early warning more than 50 days in advance, with an accuracy rate of 96.7%, and can locate the cells of subsequently develops faults. Therefore, early warning of lithium-battery-system failures is realized through unsupervised learning.

  Figures and Tables | References | Related Articles | Metrics

  Two-stroke braking performance of diesel engine equipped with variable valve mechanism

  ZHANG Xu, TAN Huiying, WU Weihai, XIE Zongfa

  2025, 16(2):  243-251.  doi:10.3969/j.issn.1674-8484.2025.02.007

  Abstract ( 7 )   HTML ( 1)   PDF (1998KB) ( 3 )  

  To address the issue of excessive driving system load caused by high valve opening pressure during two-stroke braking in heavy-duty diesel engines, this study proposes an integrated hydraulic variable mode valve system (HVMVS) capable of dynamic valve motion pattern switching. A six-cylinder diesel engine simulation model was established using GT-Power software to investigate optimal valve operation parameters that achieved high braking torque while maintaining low maximum cylinder pressure. Brake performance analysis was conducted across different rotational speeds for diesel engines equipped with HVMVS, including comparative evaluations with conventional variable valve systems. The results show that at rated engine speed, the HVMVS-equipped two-stroke braking system generates 1 158 Nm effective braking torque with a maximum cylinder pressure of 2.3 MPa, leading to a 62% reduction compared to alternative systems. The HVMVS demonstrates significant advantages in mitigating both maximum in-cylinder pressure and drive mechanism load during two-stroke braking operations, thereby enhancing operational reliability. These findings indicate promising application prospects for HVMVS in heavy-duty engine applications.

  Figures and Tables | References | Related Articles | Metrics

  Simulation and experimental analysis of proton exchange membrane fuel cell test system based on AMESim

  ZHOU Tianpeng, NIU Limin, YIN Jianhua, SI Ming

  2025, 16(2):  252-259.  doi:10.3969/j.issn.1674-8484.2025.02.008

  Abstract ( 12 )   HTML ( 2)   PDF (1470KB) ( 3 )  

  In order to assist the development of the fuel cell test system and the regulation of operating parameters, it is necessary to construct a system simulation model to optimize the matching of key parameters and analyze and evaluate the performance of the system, so as to formulate the control strategy of the fuel cell test system and improve the performance and reliability of the system. The key components were determined according to the topological structure of the fuel cell stack test bench, component models were established and simulation models of the fuel cell test system were built, and parameters of each component in the system model were calibrated according to the experimental requirements. The output results of key performance parameters in the test system were predicted and evaluated by simulation, and the simulation results were compared with the actual test data. The results show that the experimental data of the fuel cell test system is in high agreement with the simulation results of the established model, and the calculated average absolute percentage error of each parameter index of the fuel cell is 3.65% at most. The model can be used for the performance research of the fuel cell test bench and the optimization of control strategy. It provides an effective simulation tool support for developing test system and improving dynamic response performance of fuel cell.

  Figures and Tables | References | Related Articles | Metrics

  Influence of the diffusion layer surface structure on droplet flow characteristics in the PEMFC with a baffle flow channel

  SONG Jiaojiao, WANG Han, ZHANG Xiaojian, YANG Ruoxi, CAO Bo, WANG Yulin, ZHANG Jianbo

  2025, 16(2):  260-267.  doi:10.3969/j.issn.1674-8484.2025.02.009

  Abstract ( 10 )   HTML ( 3)   PDF (2136KB) ( 7 )  

  In order to improve the transport characteristics of liquid water inside a proton exchange membrane fuel cell (PEMFC), the surface structure characteristics of the gas diffusion layer (GDL) were obtained by white light interferometry and the seated drop method, and the influence of the surface structure of the diffusion layer inside the baffle-type flow channel on the flow characteristics of the droplets was investigated by the volume-of-fluid (VOF) method. The results show that, regardless of whether the surface structure of GDL is considered or not, the hydrophobicity of the surface is significantly enhanced with the increase of polytetrafluoroethylene (PTFE) content in GDL, which leads to the shortening of the outflow time of droplets in the baffle-type runner, the reduction of the spreading area of the droplets, and the corresponding reduction of the pressure drop in the runner. And compared with the smooth GDL surface, the irregular structure and pore characteristics of the actual rough GDL surface significantly enhance the droplet adhesion, resulting in a longer flow time of the droplets on the rough surface, the spreading area increasing, and the pressure drop in the flow channel increasing. When the PTFE mass percentage content is 40%, the droplet flow time on the actual rough GDL surface is shortened by 28.57%, the pressure drop in the flow channel is reduced by 16.32%, and the spreading area of the droplets showes a tendency to decrease.

  Figures and Tables | References | Related Articles | Metrics

  Predictive energy-saving technologies of hybrid electric vehicles and an evaluation method

  YANG Jianjun, LIU Dongwei, LI Jingyuan, LIU Shaohui, ZHANG Feilong, WANG Mengyuan

  2025, 16(2):  268-276.  doi:10.3969/j.issn.1674-8484.2025.02.010

  Abstract ( 8 )   HTML ( 1)   PDF (1549KB) ( 4 )  

  A standardized test-evaluation method was proposed for the hybrid electric vehicles with or without “plug-in” to investigate the predictive energy saving technologies for the national standards pre-research for automobile energy saving. By integrating map traffic information into road foresight technology, a set of critical data fields required for predictive energy saving was identified as well as the necessity of standardization was emphasized. The energy-saving mechanism was improved through dynamic optimization: the state of charge (SOC) reference trajectory was computed using global dynamic programming, and tracking control was achieved via PI control. Simulation analyzed influence of initial SOC (20%~30%) and test conditions on the energy-saving performance. Based on these findings, a dynamometer test method was developed to standardize evaluation processes. An evaluation was conducted on the energy-saving effects of dynamometer tests for a specific vehicle. The results indicate that the dynamometer test on the vehicle achieved a 3.97% energy saving, with the SOC variation trend aligning with theoretical expectations. This demonstrates the feasibility of evaluating energy-saving performance through dynamometer testing methods.

  Figures and Tables | References | Related Articles | Metrics

  Deterioration fault prediction of the drive-motor cooling-system for new energy vehicles

  LIU Chiwei, HUANG Yundi

  2025, 16(2):  277-285.  doi:10.3969/j.issn.1674-8484.2025.02.011

  Abstract ( 4 )   HTML ( 2)   PDF (1380KB) ( 2 )  

  A multi-classifier model of Principal-Component-Analysis and the Particle-Swarm-Optimization Support-Vector-Machine (PCA-GOA-LSSVM) was proposed to detect and predict the deterioration of the cooling system of the drive motor of new energy vehicles as early as possible and reduce the occurrence of motor power limit or shutdown caused by excessive coolant temperature. The Principal Component Analysis (PCA) method was used to reduce the dimensionality and reconstruct the fault characteristics. The Grasshopper Optimization Algorithm (GOA) was used to optimize parameters of Least Square Support Vector Machine (LSSVM). The sample data collected from the real vehicle fault test, were respectively input to the LSSVM prediction model, (PCA-PSO-SVM), and the PCA-GOA-LSSVM models for comparison testing. The results show that for the multi-classification prediction model based on PCA-GOA-LSSVM, the accuracy reaches 91.41% with a precision of 86.25%, which is higher than the compared prediction model. The model can be used in the performance deterioration prediction and fault diagnosis of the cooling system of the drive motor of new energy vehicles, and can accurately remind to maintain the vehicle timely and effectively judge the fault type.

  Figures and Tables | References | Related Articles | Metrics

  Intelligent Driving and Intelligent Transportation

  Trajectory generation algorithm for simulated vehicles based on trajectory prediction models

  WANG Zhenyu, YU Zhuoping, TIAN Wei, XIONG Lu, LI Zhuoren

  2025, 16(2):  286-293.  doi:10.3969/j.issn.1674-8484.2025.02.012

  Abstract ( 5 )   HTML ( 1)   PDF (1826KB) ( 3 )  

  To enhance the overall realism of background interactive vehicle trajectories in digital simulation scenarios for autonomous driving, this study approached the problem from both microscopic and macroscopic perspectives. Firstly, vehicle trajectory prediction models were trained on naturalistic driving data. Leveraging the characteristic that model-predicted trajectories more closely resembled real-world vehicle trajectories, the prediction served as the artificial intelligence (AI) driver model for background vehicles in simulation environments, improving the microscopic realism of simulated vehicle trajectory interactions. Building on this foundation, a measurement method for trajectory feature parameter statistical distribution differences and a corresponding optimization algorithm were designed, to re-select a single trajectory with the highest probability from multiple multi-modal prediction outputs, as the final driving trajectory for simulated vehicles, further enhancing the macroscopic realism of the generated trajectory feature parameter statistical distribution. The results show that, based on the proposed measurement metrics, the distribution difference between optimized simulated trajectories and real trajectories is reduced by 56.29% compared to pre-optimization, effectively enhancing the realism of background vehicle trajectories in simulation scenarios.

  Figures and Tables | References | Related Articles | Metrics

  Identification of battery polarization parameters based on initial charging segment of cloud data

  WANG Limei, CUI Yanwei, SUN Jingjing, ZHAO Xiuliang, LIU Liang, PAN Chaofeng

  2025, 16(2):  294-302.  doi:10.3969/j.issn.1674-8484.2025.02.013

  Abstract ( 8 )   HTML ( 1)   PDF (1776KB) ( 1 )  

  A benchmark polarization parameter identification method was proposed based on cloud data to enhance the accuracy and the speed of online identification of battery polarization parameters. The characteristics of battery polarization parameters were investigated by conducting charge-discharge pulse experiments. A method analogous was employed by utilizing the initial charging segment from cloud data through the Hybrid Pulse Power Characterization (HPPC) tests to obtain the charging polarization parameters. The Variable Forgetting Factor Recursive Least Squares (VFFRLS) algorithm was applied with the identified charging polarization parameters as constraints to compute the discharging polarization parameters. The results indicated that this method yielded battery time constants ranging from 34~53 s, and the polarization parameters remained invariant with respect to the current rate under corresponding low current rates in the cloud environment. The calculated charging polarization resistance and polarization capacitance aligned well with laboratory results. The convergence speed of the proposed constrained online identification method was improved by at least 6% compared with the unconstrained identification method.

  Figures and Tables | References | Related Articles | Metrics

  Research on adaptive trajectory tracking control method for intelligent vehicle

  ZHANG Shuo, LI Xiao, CHEN Yisong, ZHAO Xuan, YU Qiang, YU Man

  2025, 16(2):  303-314.  doi:10.3969/j.issn.1674-8484.2025.02.014

  Abstract ( 6 )   HTML ( 2)   PDF (4795KB) ( 3 )  

  Aiming at the problem of poor trajectory tracking accuracy and handling stability of intelligent vehicles under variable speed and variable road adhesion coefficient conditions, an adaptive trajectory tracking control method based on model predictive control (MPC) was designed. Based on the lateral force sliding mode observer and the inverse model of magic tire, the tire equivalent cornering stiffness estimation method was designed to correct the dynamic model parameters in real time. A dynamic predictive time-domain control strategy that took into account the road adhesion coefficient and driving speed was developed, and an adaptive MPC trajectory tracking controller was established. The effectiveness of the adaptive model predictive control method was verified by Simulink-CarSim joint simulation under the conditions of double lane change with variable speed and road adhesion coefficient compared with the traditional MPC control method. The results show that compared with the traditional MPC control method, the control stability of the proposed method is improved at high speed and variable speed on the high adhesion coefficient road, and the average yaw speed is improved by 19.82% at a slight sacrifice of tracking accuracy. The average lateral offset and yaw velocity are reduced by 84.90% and 46.23% respectively when driving at medium and low speed on the road surface with variable adhesion coefficient, which can effectively improve the trajectory tracking control accuracy and handling stability.

  Figures and Tables | References | Related Articles | Metrics

  Research on vehicle integrated navigation system based on improved sample convolutional interaction network

  KUANG Xinghong, YAN Biyun

  2025, 16(2):  315-325.  doi:10.3969/j.issn.1674-8484.2025.02.015

  Abstract ( 3 )   HTML ( 1)   PDF (3976KB) ( 1 )  

  Global Navigation Satellite System / Inertial Navigation System (GNSS/INS) integrated navigation system in vehicles is prone to signal loss in obstructed environments, leading to divergent positioning results and compromising the efficiency and safety of unmanned vehicles. To address this issue, this study proposed an artificial intelligence solution based on an improved Sample Convolution and Interaction Network (SCINet), which incorporated strategies such as principal component analysis, trend decomposition, and linear convolutional interactive learning on a low-layer SCINet architecture, enhancing the stability and accuracy of the model under such operating conditions. The results show that the proposed model reduces positioning errors by 80.9% and 67.6% compared to Long Short-Term Memory (LSTM) and SCINet, respectively, effectively improving the outdoor positioning accuracy of unmanned vehicles during GNSS signal loss and ensuring the reliability and safety of unmanned vehicle positioning.

  Figures and Tables | References | Related Articles | Metrics

  Risk-sensitive hierarchical reinforcement learning decision-making for autonomous vehicles

  HU Zhilong, PEI Xiaofei, ZHOU Honglong, WEI Weiran

  2025, 16(2):  326-333.  doi:10.3969/j.issn.1674-8484.2025.02.016

  Abstract ( 6 )   HTML ( 1)   PDF (1171KB) ( 3 )  

  In order to make the behavior decision of autonomous vehicles fully consider the inherent uncertainty in the traffic environment, this paper introduced quantile regression and Conditional Value at Risk (CVaR) based on the traditional RainbowDQN algorithm, taking low-probability risks into consideration, and properly balancing risks and benefits, so that it can make safer and more humane driving decisions. A behavioral decision model was established based on the Markov framework, and the reward function and action space were designed by comprehensively considering safety, efficiency and comfort. A planning and control model was built, and two scenarios of highway inflow and outflow and intersection were built using the Open Natural Driving Intelligent Vehicle Simulation Test Environment (OnSite) platform. The OnSite evaluation tool was used to simulate and compare the four algorithms of RainbowDQN-CVaR, RainbowDQN-QR, RainbowDQN and DSAC-T. The results show that in complex highway merging and exiting scenarios and intersection scenarios, the proposed RainbowDQN-CVaR algorithm scores 55.3% and 47% higher than the traditional RainbowDQN algorithm, 17.7% and 34.3% higher than the RainbowDQN-QR algorithm, and 2.8% and 62.7% higher than the DSAC-T algorithm. The effectiveness of the RainbowDQN-CVaR behavior decision model is verified, and it can make safer and more reasonable decisions in a more complex traffic environment, making the autonomous driving vehicle have higher driving safety and efficiency.

  Figures and Tables | References | Related Articles | Metrics

  Design and research of an automated parking evaluation tool based on the OnSite platform

  YANG Junru, ZHENG Sifa, XU Shucai, TIAN Ye, SUN Jian, SUN Chuan, LI Haoran

  2025, 16(2):  334-343.  doi:10.3969/j.issn.1674-8484.2025.02.017

  Abstract ( 6 )   HTML ( 2)   PDF (2735KB) ( 7 )  

  An automated parking evaluation tool was developed to enhance the functionality of the platform OnSite (Open Naturalistic Simulation and Testing Environment) for autonomous driving. This tool used a scenario construction method based on real vehicle data collection and modeling reconstruction. A more comprehensive test scenario library was established according to industry standards and parking space data. A multidimensional evaluation system was proposed, focusing on completion rate while considering safety, efficiency, and accuracy. The evaluation tool underwent hardware-in-the-loop simulation and was compared with results from the CARLA simulation platform and real vehicle tests. Scores of the top 10 teams in the parking test of the 2nd OnSite Autonomous Driving Algorithm Challenge were analyzed to discuss the future development of the evaluation tool and the OnSite platform. The results show that the tool provides a scientific basis for optimizing automated parking functions and serves as a reference for developing autonomous driving evaluation tools.

  Figures and Tables | References | Related Articles | Metrics