The development of visual perception technology based on deep learning is beneficial for the advancement of environment perception technology in automatic driving systems. However， for corner cases of autonomous driving scenario， there are still some problems in the current perception model. This is because the ability of the perception model based on deep learning depends on the distribution of the training dataset. Especially when categories in the driving scene never appear in the training set， the perception system is often fragile. Therefore， identifying unknown categories and extreme scenarios remains a challenge for the safety of automatic driving perception technology. From the perspective of processing data sets， in this paper a novel multimodal automatic corner case mining process called "Corner Case Mining Pipeline （CCMP）" is proposed. In order to verify the effectiveness of "CCMP"， the concern case subset "Waymo-Anomaly" on the basis of Waymo open datasets is established， with a total of 3 200 images， each of which will contain the corner case scene defined in the text. Then based on the private data set Waymo-Anomaly， it is proved that the recall rate of "CCMP" corner case mining can reach 91.7%. In addition， the effectiveness of object detectors targeting long-tailed distributions in datasets containing corner case is experimentally verified. Ultimately， the authenticity of the automatic driving perception model in the real world is expected to improve from the perspective of datasets processing.

Accurate performance evaluation of power battery cells is of great significance to ensuring the safety of power batteries. For the existing data-driven battery fault diagnosis algorithms， mostly individual cells are compared with each other and the outlier cells are identified as faulty cells by classification， based on differences in characteristic parameters such as single cell voltage. However， if there are multiple cells of similar abnormally performance in the power battery pack， or all individual batteries show an overall performance deterioration， it is difficult to distinguish individual cells or even there is no significant outliers， and the application of the mutual comparison strategy is limited. A power battery fault diagnosis method is proposed based on 1dCNN-LSTM to quantify the abnormality of a single cell in this paper. Combining the three types of characteristics of vehicle motion status， drive system status and power battery electrical signal， the 1dCNN-LSTM fusion model is established to estimate the individual cell voltage under ideal conditions as reference. The difference between the real-time voltage reference value and the measured voltage value is used to quantify the abnormality of each cell. Combined with actual cases， it is shown that for thermal runaway case due to single cell failure， the abnormal performance of the faulty cell compared to others can be identified 7 days before accident， and potential risk can be recognized in discharge processes from a year of more before the accident. For overall deterioration cases without obvious individual cells inconsistency， the deterioration evolution within the last 7 days can be tracked.

To address limitation in prediction accuracy and data utilization efficiency of supervised learning-based trajectory prediction models， a trajectory prediction model and a general self-supervised pretraining strategy are proposed. Firstly， a lightweight trajectory prediction model based on Transformer is established to extract temporal-spatial features while modeling interaction relationship. Secondly， three types of masks， namely motion information temporal mask， road information spatial mask， and interaction relationship mask， are designed for self-supervised pre-training tasks on the model to enhance the model's ability to extract general scene features. Finally， pretraining weights are used as initialization parameters for supervised learning fine-tuning in downstream tasks. Experimental results on the Argoverse2 Motion Forecasting dataset show that the model can effectively reconstruct traffic scenes in pretraining tasks. The introduction of self-supervised pretraining improves prediction accuracy and data utilization efficiency. Moreover， it exhibits universality for different prediction tasks， achieving a 3.3% and 3.7% improvement in the minFDE₆ for single-agent and multi-agent trajectory prediction tasks， respectively.

Most of the current vehicle route planning is based on the grid map planning method， which will greatly increase the amount of calculation when the search area is large. In contrast， the method based on visibility graph can reduce the amount of calculation during path search， but is greatly affected by the complexity of obstacles. For this problem， combining the SLAM and visibility graph methods， a simplified visibility graph construction and planning method is proposed in this paper. Firstly， the improved SLAM algorithm is used to generate point cloud maps， and dynamic obstacles are removed. Then a visibility graph is generated， and the complex edges of polygons in the visibility graph are simplified based on the size of the obstacle and the size of the concave angle at the vertex to eliminate redundant vertices. Finally， through simulation experiments and real vehicle experiments， it is proved that compared with the original algorithm， this method can reduce the number of polygon vertices in the visibility graph by 20%-30% while ensuring the accuracy of mapping. The map update time and the running time of the overall algorithm are also reduced by more than 30%. It shows that the method in this paper can effectively reduce the amount of calculation and the running time of the algorithm in the mapping and planning process.

In recent years， with the continuous enhancement of fuel cell power， hydrogen supply systems have been evolving towards blind-end anode topologies with hydrogen circulation. However， research on testing systems for hydrogen supply and circulation lags noticeably， particularly in the performance testing of core components such as hydrogen circulation pumps and injectors. Therefore， a multifunctional testing platform for fuel cell hydrogen supply systems is developed in this paper， enabling component testing， characteristic data acquisition， offline calibration， and other functionalities for hydrogen circulation systems with different configurations. The platform， by simulating the pressure drop， hydrogen consumption， and the production of water and heat in real fuel cells， mitigates the additional cost incurred by testing on the performance and lifespan of actual fuel cells. Ultimately， based on this platform， an anode pressure control and anode purge control test are conducted on the hydrogen supply system of a 150 kW fuel cell， verifying the capability of the developed testing platform to meet specific testing requirements for different loads.

The requirement of low carbon and lightweight in the auto industry is growing now. The new mega-casting technology applied on vehicle body can better achieve weight， cost and emission reduction， and has become spotlight to automobile manufacturers. In this paper， the traditional steel front compartment of passenger car body is replaced by integrated die casting part， and lightweight design on the aluminum alloy integrated front engine compartment is conducted. The optimal load path for stiffness is obtained through topology optimization of the front cabin by SIMP method. Considering the castability of the front cabin， the draft direction， thickness size and position distribution of the ribs are designed. Frontal impact simulation is conducted according to C-NCAP2021 and the impact resistance of the integrated die cast body is improved through Taguchi experimental design method and response surface optimization. Simulation analysis is conducted on the optimized performance of the white body. Compared with the traditional scheme， the weight of the optimal design is reduced by 13.9%， with the bending stiffness of the BIW increased by 9.7%， and the first modal meets the requirements. The research in this paper is meaningful for the platform design and industrialized application of integrated die casting car body structure in the future.

The heat exchange effect of internal compressed air leads to strong thermal hysteresis and frequency correlation of air springs’ mechanical properties. Therefore， a thermal hysteresis equivalent mechanical model is constructed to describe the energy exchange process of compressed air inside air springs in this paper. Based on the rubber airbag modal， an air spring hysteresis mechanical characteristic model covering both rubber airbag hysteresis and compressed air thermal hysteresis is constructed， and an identification method for the key parameters of the model is provided. The experiments show that the maximum errors of the hysteresis loop and dynamic stiffness are less than 3.3% and 6.7%， respectively， verifying the accuracy of the hysteresis mechanical characteristic model. Finally， the inherent law of the thermal hysteresis of compressed air with frequency varying is revealed. The research results provide theoretical support for identifying the hysteresis nonlinear mechanism of air springs and its effective utilization.

Aluminum alloy multi-cell thin-walled tubes have better mechanical properties in energy absorption than ordinary square tubes in axial compression conditions， with a wide range of application prospects in automotive， aviation， military equipment， and other industries. To study the anisotropic characteristics of extruded 6061-T6 aluminum alloy material， uniaxial tensile mechanical properties tests are conducted on the sheet along the extrusion direction of 0°， 45°， and 90°. The corresponding stress-strain curves and anisotropic characteristic parameters are obtained， and the material constitutive model is established based on the yield criterion of anisotropic hardening behavior. Tubes with different cross-sectional configurations shaped as the Chinese characters of mouth， day and eye are designed and quasi-static crushing tests are conducted. By analyzing the deformation crushing force curve， it is shown that the thin-walled structure of the triple-cell alloy has superior crash resistance performance. In order to further obtain the optimal design parameters of the triple-shaped tube， considering the uncertain effect of material parameter fluctuations such as Poisson's ratio and elastic modulus on the structural impact resistance， the multi-cell aluminum alloy thin-walled tube impact resistance interval uncertainty optimization model is established. The interval possibility degree method is used to transform it into a deterministic problem. By combining the Artificial Neural Networks （ANNs） model with the Intergeneration Projection Genetic Algorithm （IP-GA） method， a double-layer nested optimization is performed on this problem to analyze the impact of different likelihood levels on uncertainty optimization results， providing guidance for the selection of different reliability optimization design.

Electrically controlled air suspension （ECAS） has the function of adjusting suspension stiffness and body height， which can effectively improve vehicle ride comfort and handling stability. Taking a passenger car ECAS as an example， the viscoelastic damping characteristics of rubber airbag are described by fractional theory， and the thermodynamic model is optimized considering the equivalent damping and hysteretic characteristics， which is in good agreement with the experimental data， and the precision of the optimized air spring model is verified. On this basis， considering the longitudinal and lateral dynamic characteristics of the vehicle and the Dugoff tire model， a 14-degree-of-freedom vehicle ECAS dynamic model is established， and a Model Predictive Control （MPC） active suspension control method is proposed， with measurable variables as the input of the controller， to realize the active control under straight and turning driving conditions. Simulation and vehicle bench test show that the fractional correction model can well reflect the variable stiffness characteristics of ECAS， and the active suspension control strategy based on MPC can adjust the air spring stiffness in real time， control the body posture， and effectively improve the ride comfort and stability of the electric vehicle. The research method in this paper provides a new idea for vehicle suspension system modeling and active control.

The promotion of intelligent cockpit and virtual testing protocols bring new challenge to assess the occupant injury，with the injury mechanism and injury risk assessment parameters more diversified. Based on the TUST IBMs 6YO-O and the BP neural network algorithm，a predictive model for the correlation between occupant sitting angle and head injury indicators in frontal 100% overlapping rigid barrier condition is constructed in this paper，and the correlation and difference between evaluation indicators with the different seating postures are explored. The results show that the constructed correlation injury prediction model has high reliabilities （R ² > 0.90），which can be used for injury prediction and analysis. Existing head injury evaluation indicators have good consistency in the small angle range （95°~108°），but for the occupants with larger seating postures，there are significant differences to assess the head injury risks using different injury evaluation indicators. Therefore，there is certain limitation of the head injury assessment parameters implemented currently. In the future virtual testing，the kinematic and biomechanical parameters should be integrated to assess more comprehensively for the head injury risks. The research results can provide data and theoretical support for the improvement of child restraint systems，virtual testing，and selection of head injury evaluation parameters for occupants with larger seating postures.