In order to realize active equalization of 48 V mild hybrid battery pack, and improve energy utilization rate, firstly the group design and consistency check of 48 V mild hybrid system battery of a vehicle were conducted, then the topological structure of the active equalization circuit was determined, and then the equalization opening threshold and equalization control strategy of active equalization circuit were designed based on the information of the battery terminal voltage collected by the BMS. The results of simulation analysis indicate that the State Of Charge (SOC) bias errors of the battery module after active equalization does not exceed 0.005.

To address the issue of the unsuitability of the original suspension parameters after upgrading a central motor electric vehicle to a In Wheel motor car, A half-car passive suspension model is established in MATLAB/Simulink to analyze the problem. The spring stiffness and damping coefficients of the front and rear suspensions are taken as independent variables, while the root mean square values of the vehicle’s center of gravity acceleration, pitch angle acceleration, suspension deflection, and tire load are considered as optimization objectives. The grey wolf optimization algorithm is employed to search for the optimal stiffness and damping parameters. The simulation results show that after optimization, the handling stability of the suspension remains within a reasonable range, while the comfort level is improved by 12.73%.

To improve the low efficiency of the current diagnosis method using subjective evaluation method, this paper proposed a method for identifying vehicle abnormal noise based on Convolutional Neural Network (CNN). A four post vibration test stand was used to collect raw abnormal noise signals with high signal-to-noise ratio in a semi anechoic chamber environment as the research object to extract the Mel-spectrogram of the signal as input to the neural network, then, a convolutional neural network was constructed to perform deep level feature extraction, compression, and classification recognition on the data. It was found that the average recognition rate can reach 90.5%. Finally, the transfer learning method was used to optimize the model. The test results indicate that VGG and ResNet models can improve recognition accuracy, and the ResNet network has better recognition performance in the test set.

A testing device for air pressure resistance of door seal has been designed to simulate the changes of air pressure in door cavity of a car under side crash condition and test the ultimate air pressure resistance of door seal. Based on this, this article studied the influence of different temperatures, material hardness, and part structure on air pressure resistance of the door seal in combination with a high and low temperature environment box. Through multiple experimental comparisons, it has been proven that the ultimate air pressure resistance of door seal is negatively correlated with temperature. Increasing the back plate ribs and appropriately increasing the material hardness can significantly improve air pressure resistance.

In order to deeply explore Yolo target detection algorithm and its development direction in the field of autonomous driving, this paper reviewed the research progress and main accomplishments of YOLO algorithms (YOLOv1~YOLOv8), analyzed and expounded its application in relevant detection tasks including vehicles, lane lines, pedestrians and traffic signs in the field of autonomous driving. Combining the characteristics of Yolo series algorithms, the paper analyzed and summarized the future improvement ideas and research trends of Yolo algorithm.

In view of the problem that the variable window Adaptive Unscented Kalman Filter (AUKF) algorithm has a large mutation when the window changes, and the window sequence data decreases sharply, resulting in the increase of error, stability and accuracy decline in state estimation, this paper uses the Forgetting Factor Recursive Least Square (FFRLS) algorithm to identify parameters based on the second-order RC equivalent circuit model, combined with the improved variable window AUKF algorithm to estimate the State of Charge (SOC) of lithium battery, it is verified by the Urban Dynamometer Driving Schedule (UDDS) cycle test, and compared with Unscented Kalman Filter (UKF), AUKF and variable window AUKF algorithms. The test results show that the improved AUKF algorithm can control the average error within 0.38%, with higher accuracy and convergence.

In the road test, an electric vehicle was found brake stall under high SOC, and paused under low SOC when it was running at full throttle, it was found and confirmed that irrational current signal parameters injected by the harmonics of the inverter software triggered resonance of the vehicle electrical network, which caused voltage jump. The resonance was analyzed, and a vehicle electrical network model was established, and the resonance point was calculated, the harmonic injection strategy was optimized. Vehicle test shows that under the new strategy, the voltage jump has been reduced and it does not cause new NVH problems.

This paper took noise generated from air suspension modulation of a vehicle as research object. Through transmission path study, the paper determined that the noise was mainly generated through the body structure excited by the air compressor. Through theoretical analysis, it was found that the main factors affecting air suspension noise transmission were the air compressor system modal frequency and Noise Transfer Function (NTF) from the body attaching point to the vehicle interior. The paper proposed optimization schemes including power supply wiring harness wrapping material optimization, bracket vibration isolation and bracket reinforcement + NTF. The results show that bracket vibration isolation is the best scheme, after implementation, the 1^st order noise order of air compressor was reduced by 4 dB, 2^nd order noise was reduced by 14 dB, and the sound quality in the vehicle was significantly improved.

In order to solve the problem of abnormal vibration and noise generated by the Electric Power Steering (EPS) system of passenger cars under bumpy road conditions, this paper proposed a new damped block structure by adding an O-ring at the EPS block position. By analyzing the influence of O-ring static compression and stiffness on damping performance, the optimal compression of O-ring was confirmed to be 0.5 mm. The EPS endurance test was carried out with the road spectrum of the vehicle. The results show that the performance of vibration noise deteriorates after a long mileage with increase of EPS block gap. Subsequently, the bench noise test of the EPS with traditional block and new block under different block gap was conducted. The results show that the vibration acceleration of the EPS with the new damped block structure is reduced by a maximum of 9 m/s² and 6.2 m/s² at the bushing end and block end, respectively, proving that the new damped block structure has a significant optimization effect on the vibration noise of the EPS system under bumpy road conditions.

To obtain engine braking power accurately, this paper proposed a method of engine braking road test based on the momentum conservation law. A vehicle braking power equation based on speed and gear was built to provide a warning of downhill shifts and AT gear selection. Braking power test result obtained with the proposed test method was compared with dynamometer test result, with a difference range of -2.7%~3.4%, demonstrating accuracy and effectiveness of this method.