In this paper, an electric vehicle's aerodynamic drag and wake are numerically studied. The results show that the flow separates from the rear of the car may roll up into a largescale vortex at Re_L=1.1 × 10⁷. The ratio of the RMS drag and the mean drag reaches to 3.27%, making an unneglectable effect on ride comfort and mileage prediction. The pressure at the back, the underbody, the middle and lower parts of the near wake, the aerodynamic resistance of the entire vehicle, the pressure near the wall of the rear guard plate in the bottom, and the separation flow at the bottom all have a characteristic frequency of 12 Hz. However, the flow separation at the top and Cpillar of the car does not have this characteristic frequency. It proves that the underbody flow separation at the rear is the main cause of dynamic changes of the aerodynamic drag.

In order to adapt to the high power density of automotive high-speed motors and the high thermal load under extreme working conditions, the current motor cooling mostly adopts the direct contact oil cooling heat dissipation method, and it is necessary to establish a motor oil temperature model suitable for the study of thermal control methods. Existing methods are mainly based on finite element simulation calibration, which cannot meet the realtime application requirements, while the multi-physical field coupling of the complex oilwater heat transfer circuit makes it difficult for the online reconstruction of oil temperature. In this paper, a secondorder lumpedparameter oil temperature model is proposed to strengthen the time-sequence cyclic process and consider the strong autocor-relation. The oil circuit unit is modeled according to the calibration, and the motor loss response is determined based on bench-top measurements. The time-sequence convolution method is adopted to describe the heat transfer process, and a cyclic dynamic recursive model with high and low oil temperature coupling is established. Oil temper-ature-sensitive parameters are introduced to improve the adaptability of the working conditions to solve the difficult problem of describing the oil temperature distribution in the flow path. Finally, the model accuracy is verified online by road spectrum working conditions, with the average absolute estimation error of the oil coolant temperature within 1°C, which can support the refined thermal management of the motor.

To meet the stringent thermal management requirements of electric vehicles and address global climate change issues, in this paper an integrated thermal management system is developed for electric vehicles based on R290 refrigerant. The system's performance is analyzed and validated through simulation and experiments. The results show that the cooling and heating capacities of the R290 dualside indirect thermal management system increase with the increase of the compressor speed, while the coefficient of performance (COP) decreases with the increase of the compressor speed. The heating capacity from the compressor's hot gas bypass increases with higher system pressures. Under hightemperature cooling conditions at 40 °C, the system's maximum cooling capacity is 9.25 kW. Under lowtemperature heating conditions at 18 °C, the maximum heating capacity is 7.24 kW. At extremely low temperatures of 20 °C, the maximum heating capacity from the compressor's hot gas bypass is 4.3 kW.

A noninvasive driving fatigue state identification method based on the surface electromyographic signals of the driver's legs is proposed. Firstly, the electromyographic signal of the tibialis anterior muscle of the driver's right leg is collected through a simulated driving fatigue experiment, and the fatigue status is marked through a subjective evaluation scale. Secondly, a variational mode decomposition algorithm is used to filter out noise on the surface electromyographic signal, and 12 timefrequency domain eigenvalues are extracted from the five IMF components obtained by decomposition. Finally, a driving fatigue state discrimination model based on whale algorithm optimized support vector machine is constructed. The results show that this method has a good discrimination effect on three fatigue states, with an accuracy of more than 84%.

Developing a highprecision Statistical Energy Analysis (SEA) model to predict vehicle wind noise response requires a significant amount of time and cost. In this paper, a method is proposed for rapidly constructing an equivalent SEA model for vehicle wind noise based on parameter identification, which simplifies the modeling process while ensuring prediction accuracy. An initial SEA model of the compartment is established according to the vehicle's body structure and dimensions, with the pressure fluctuation excitation on the side window surface and the actual wind tunnel response serving as the model's input and output, respectively. The Grey Wolf Optimizer (GWO) algorithm is employed to identify the acoustic cavity parameters of the model, resulting in an equivalent model that approximates the true wind noise response characteristics. Taking a prototype vehicle as an example, the equivalent wind noise SEA model is used to predict the wind noise response in the compartment under different design schemes. The average prediction error for the total sound pressure level is 1.47%, and the root mean square error of the spectrum is 1.23 dB. The results show that the equivalent model can accurately predict the invehicle wind noise response under different design schemes, thereby reducing the number of wind tunnel tests and having high engineering application value.

In order to solve the problem of roaring sound inside the vehicle caused by intermittent engine intervention during charging and discharging of dieselelectric hybrid vehicles, in this paper a semicoupled cluster control strategy with better comprehensive performance is proposed based on the traditional multichannel active noise control (ANC) system by combining the advantages of the centralized control strategy and decentralized control strategy. Compared with the centralized control strategy, the computational cost of the cluster control strategy is reduced by about 50%, and the noise attenuation performance is comparable to that of the centralized control strategy. Compared with the decentralized control strategy, the stability is obviously better, and the noise reduction effect is outstanding. Based on the MATLAB simulation platform, a variety of cluster control strategies and traditional control strategies in the vehicle are compared and analyzed, and the road test experiments of a rangeextended electric vehicle are carried out under its common working conditions. The results show that the cluster control strategy can be well applied to the multichannel active noise control system in the vehicle, and the average noise reduction amount of the second, fourth, and sixthorder range extender noise at the four seat headrest positions can reach 15.9, 10.6 and 5.7 dB(A), respectively, showing good noise reduction effect and stability. The research results can be applied to the noise control of manned cabins, such as aircraft, submarines and other fields, which has important scientific significance and engineering value.

The total vehicle mass is an important parameter for both power and safety control of the vehicle, especially for heavyduty trucks. Based on the theory of vehicle longitudinal dynamics, a method for estimating the total vehicle mass according to the vehicle operating conditions is proposed in this paper. Firstly, under acceleration conditions, engine torque and longitudinal acceleration are obtained through CAN bus, using the Kalman filter algorithm (KF) to estimate the total vehicle mass. Then the estimated mass is used to identify the unknown parameters. At constant speed, the mass is estimated based on the identified unknown parameters and a simplified vehicle longitudinal dynamics model. The effectiveness of the method is verified through joint simulation suing the estimator constructed by TruckSim/Simulink. By the vehicle road test, the results show that the method is able to estimate the total vehicle mass more accurately under different operating conditions.

An optimization method incorporating the Improved Harris Hawks Optimization (IHHO) algorithm is proposed for the lightweight research of a trusstype snowplow frame. Firstly, a finite element simulation model of the frame is constructed, and its strength, stiffness, and modal characteristics are quantitatively analyzed under various working conditions to determine its strength performance, stiffness performance, and natural frequencies. Subsequently, the Response Surface Methodology is employed, using maximum deformation and maximum stress as response variables, to optimize the crosssectional dimensions of the frame beams, yielding three sets of optimal solutions. On this foundation, the IHHO algorithm is proposed by improving the HHO algorithm, and the effectiveness of the optimal solutions is verified using the IHHO algorithm. The optimization results show that the overall mass of the frame is reduced by 33.6%, with the maximum deformation decreased by 6.33%, the maximum stress increased by 3.01%, and the firstorder modal frequency decreased by 19.48%, effectively avoiding the resonance range. This study provides an efficient and feasible optimization strategy for the lightweight design of trusstype frames. The method demonstrates significant advantages in model construction and obtaining accurate estimation results, offering theoretical references for engineering application in related fields.

In this paper a simulation testing method for intelligent vehicle based on a large language model is proposed to address the issues of heavy reliance on human resources and prominent efficiency bottlenecks in existing scenario based testing methods. Firstly, a simulation testing architecture for intelligent vehicle based on a large language model is designed, and corresponding data and simulation layers are established. On this basis, an intelligent car simulation testing process based on a large language model is constructed. Knowledge mining, model finetuning, and knowledge base enhancement retrieval application processes are designed for knowledge question answering tasks. Application paths for scenario type analysis, scenario element generation, and scenario toolchain invocation are designed for scenario generation tasks. For testing and evaluation tasks, a comprehensive application framework for testing scenario analysis, evaluation system construction, and simulation testing execution is designed. Finally, each task is tested. The results show that the testing method proposed in this paper can effectively solve different types of testing tasks and improve testing efficiency.

In the application of vehicleroad cooperative technology for dynamic display of the roadside twin maps, due to the delay problem of the communication between networked devices and the existence of the roadside perception error, the fusion perception accuracy of the roadside edge computing unit will be seriously affected, which will lead to the jitter and delay of the vehicle display track in the twin map. Hence, in this paper a vehicleroad cooperative sensing and localization method that fuses high definition map under high communication delay is proposed. The method first analyzes and models the communication delay between the vehicle end of the connected vehicle and the roadside edge processing unit in the vehicleroad cooperative system, divides the delay model into sensor synchronization delay and communication transmission delay, and proposes a synchronization optimization method for the delay. After the synchronization optimization, a collaborative multidimensional particle filter algorithm for swarm vehicles is proposed, where the states of the particles represent the pose of different connected vehicles and nonconnected vehicles in the swarm vehicles. In the proposed multidimensional particle filter algorithm, the state of the particles is firstly updated using the observation of the state of the particles by utilizing the roadside RSU observation data and the curvature information of the lanes in the highdefinition map. Then the selflocalization information of the received delayed synchronized smart connected cars combined with the left and right lane line lateral constraint information and the lane line equations of the lanes in the high definition map are used to update the observation of the state portion of the particle that represents the smart connected cars. The experimental results show that the perceptual and localization accuracy of the edge server is improved by 59.4% in the low delay scenario with less communication interference, and its accuracy is improved by 38.6% in the high delay scenario with severe communication interference. Therefore, the proposed vehicleroad cooperative sensing method incorporating high definition map under high communication delay can effectively deal with the communication delay problem and improve the multivehicle perception accuracy of the edge computing unit, thus improving the accuracy, stability and continuity of the twin map dynamic data.