In order to give full pay to the advantages of rapid torque response and independent controllable torque of the distributed driven electric vehicle, this paper adopted the chassis stability coordination control method, through the coordinated control of the electronic stability control system and the torque vector control unit to deal with the stability requirements of the vehicle under various working conditions-improve the dynamic performance of the vehicle under stable conditions through TVCU independent control, ensures the power and stability of the vehicle by ESC when it tends to be unstable. The robustness of the stability coordination control method was verified based on a real vehicle.

To address the issue of deviation in the estimation of vehicle mass and road gradient by applying the longitudinal dynamics model when there is a lateral component force in the driving process of vehicles, this paper proposes a coupled mass estimation model based on longitudinal-horizontal dynamics and an algorithm for slope estimation. By analyzing the effect of the acceleration phase on mass estimation, the mass estimation trigger condition is set and the recursive least squares method with forgetting factor is used to estimate the vehicle mass, and the kinematic Kalman filter is fused with the kinematic extended Kalman filter to jointly estimate the road slope. The algorithm is validated by Simulink-CarSim joint simulation and real vehicle test. The results show that the error of the mass estimation algorithm based on longitudinal-transverse dynamics is 0.82%, and the error of the fusion slope estimation algorithm is within 3%, which verifies that the algorithm has good accuracy and real-time performance.

In the semantics mapping construction of Simultaneous Localization And Mapping (SLAM) algorithm, to solve the problem that the semantics information extracted by using deep learning technology contains noise, this paper proposes a representation method for semantic observation by probabilistic grid. The influence of noise is reduced by grid matching and probability updating, in this way, a clear map can be incrementally constructed without auxiliary means like features matching. The effectiveness of the proposed method is verified by experiments in which clear zebra crossings and lane lines are contained in the constructed map.

To establish an effective electromagnetic radiation simulation prediction and design capability during the vehicle design phase, this paper investigates the generation mechanism of electromagnetic radiation from power cables in the motor drive system based on electromagnetic wave radiation theory, and proposes a vehicle finite element model for electromagnetic compatibility simulation. An electromagnetic interference model is built to simulate and predict the electromagnetic radiation emitted by the power cables. The accuracy of the model is validated by comparing the simulation results with actual measurement data. Utilizing this simulation model, the cable layout is optimized, thereby reducing the intensity of interior electromagnetic radiation. The findings indicate that this approach can identify the electromagnetic radiation risks associated with high-voltage cable routing during the design phase, thus avoiding costly modifications in the prototype testing stage. Furthermore, this method contributes to shortening the vehicle design and development cycle while reducing testing costs.

Based on a 200 kW peak power three-stack common-rail Proton Exchange Membrane Fuel Cell (PEMFC), the steady Computational Fluid Dynamics (CFD) iterative calculation between the single pulse valve ejector and the multi-stack common-rail pulse valve ejector under and hydrogen supply system was carried out by Fluent software, and the ejector scheme was determined. Then the reliability of the design and simulation was verified by the ejector bench test. The influence factors of pulse valve ejector were simulated and analyzed, from which we can know that with the same pressure variation, the injection pressure of secondary flow has the greatest influence on the ejection coefficient, followed by the inlet pressure, and the injection pressure has the least influence. Meanwhile, nitrogen content has great influence on the ejection coefficient. Finally, the transient simulation analysis of the pulse valve ejector was carried out. The results show that, compared with the steady state results, the ejection coefficient calculated under transient state has a large margin, so it can further cover the low power condition. Keeping the duty cycle unchanged and increasing injection frequency are beneficial to the improvement of ejection coefficient.

In order to realize the development of vehicle functional safety based on Service Oriented Architecture (SOA), the forward development process for automotive functional safety based on SOA is established combining with the ISO 26262 standard. The functional safety design methods of function-based concept phase and architecture-based system development are proposed. On the one hand, for the concept development phase, the functional safety requirements are derived based on the product capabilities, and the function safety class of product capabilities is defined. On the other hand, for the system development phase, the technical safety requirements are derived and distributed based on the software component architecture and physical deployment.

To shorten the development cycle of Dedicated Hybrid Transmission (DHT), the Hardware In the Loop (HIL) simulation platform was built to perform the virtual calibration of the electric control parameters for DHT. The obtained calibration dataset was compared with the conventional calibration data of real vehicle. The results show that the calibration data obtained from virtual calibration method is similar to that obtained from conventional method, can shorten vehicle development cycle effectively.

In order to recognize driving fatigue, this paper proposed a fatigue detection method based on wavelet characteristics and Long Short-Term Memory (LSTM) neural network classifier. Two kinds of EEG signals (non-fatigue and driving fatigue) were collected in the real driving environment. The EEG signals were decomposed by wavelet, and the statistical values, energy values and relative energy values of four wavelet coefficients were calculated as the characteristic data, which were used for classification training and test of the LSTM neural network. The results of experiment show that the classification performance of LSTM neural network gradually improves with the increase of the number of channels involved in constructing characteristic data. Especially, in the scheme of 14 channels, the average classification accuracy is about 96.1%.

Under high-frequency variable loads, EV electrical system is prone to causing bus voltage oscillations and instability. A small-signal model of the vehicle’s electrical system has been established, obtaining the output impedance of the system source-transformer and the input impedance of the motor load. Based on the traditional impedance ratio criterion and combined with the system’s stability margin, the stability determination principle of the vehicle’s electrical system has been derived. By using a Bode diagram for predicting system instability, research is conducted to explore the instability boundary of pure electric vehicle under high-speed conditions. The simulation verification results show that the vehicle’s maximum speed shall not exceed 188 km/h, and torque variation of hub motor will affect vehicle stability, the torque mutation of single hub motor shall not exceed 220 N·m, which is basically consistent with theoretical analysis results. The results prove that this stability determination principle can predict instability boundary of vehicle electrical system to some extent.

In order to satisfy the requirement of experiment and verification of vehicle platooning research, a vehicle with driving assistance function was designed, and an extensible and multi-scenario switchable vehicle platooning experimental platform was developed. The platooning vehicles are controlled horizontally by pure tracking algorithm based on visual perception and vertically by cooperative adaptive cruise based on millimeter wave. Hardware system architecture of the platooning experimental platform is built based on vehicles with driving assistance function, and vehicle role and function scene element from the software system are used to realize extension and multi-scenario switch of vehicle platooning. Finally, vehicle platooning experimental platform was established based on three minicars, function and performance test were conducted. The results show that this platooning experimental platform has function including vehicle queuing, vehicle platoon forming, and vehicle platoon driving steadily, with good control effect.

In addressing the issues of suboptimal solutions and poor exploration performance in narrow passages of intelligent car path planning using the Rapidly-exploring Random Tree-Connect (RRT-Connect) algorithm, this paper improves the RRT-Connect algorithm in expansion strategy and path smoothing based on an analysis of the basic principle of the RRT-Connect algorithm. Firstly, in terms of expansion strategy, a probability bias method is introduced to screen random points, and an expansion method based on artificial potential fields is used to shorten paths and reduce computation time. Secondly, regarding path smoothing, a third-order B-spline curve is introduced to optimize the path and generate a smooth path, ensuring that the path meet the dynamic characteristics of intelligent cars. Finally, the superiority of the improved RRT-Connect algorithm is demonstrated through comparative simulation. The results show that in environments with simple obstacles, complex obstacles and narrow paths, the average time and path length of the improved RRT-Connect algorithm are superior to those of the traditional RRT-Connect algorithm.

To address the issue of the increase of unsprung mass and suspension parameters needing adjustment after upgrading a central motor electric vehicle to a wheel hub motor car, a half-car passive suspension model was established in MATLAB/Simulink. The spring stiffness and damping coefficients of the front and rear suspensions were 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 dynamic load were considered as optimization objectives. The grey wolf optimization algorithm was 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%.

The large number of parameters in deep learning models for driver distraction detection makes it difficult to deploy them on embedded devices. To address this issue, this paper proposes a lightweight distracted driving detection algorithm, YOLOv8n-SGC, based on YOLOv8n. First, a lightweight backbone network, ShuffleNetV2, is constructed, and Ghost convolution is introduced to reduce the number of model parameters and computational cost, achieving model lightweighting. Second, a Convolution and Attention Fusion Module (CAFM) is added after the backbone network to fuse global and local features and improve the algorithm’s detection accuracy. The results show that the improved algorithm model has a reduction in parameters and computational cost compared to the benchmark model, a 28.67% reduction in volume, a 41.79% reduction in inference time, and an mAP increase of 1.1 percentage points.

For the problem that it is difficult to determine the influence of the opening size of automobile electric sunshade on the operating performance, this paper, by analyzing the structure of the electric sunshade and the basic motion parts in detail in the process of unrolling and winding of the sunshade, obtained the relationship between the stress state and the position of the sunshade in the motion process, thus established a mathematical expression of the system resistance and its opening size of the sunshade. The calculation data of the design sunshade is compared with the measurement data of the real sample, which show that, the calculated data of the sunshade model is basically consistent with the measured data of the real sample, verifying the effectiveness of the mathematical model of the sunshade system resistance.

In order to solve the problem of target detection failure of millimeter wave radar due to radar occlusion under complex road conditions, this paper proposes a radar occlusion detection method based on spectral estimation. Multi-channel signal accumulation is used to solve the two-dimensional spectrum of echo, adaptive CFAR detection algorithm is used to eliminate the radar target, and Spectrum estimation is used to solve the echo background power. The occlusion detection is accomplished by counting the energy distribution characteristics of the echo in a certain time interval. The experimental results show that the proposed method can achieve the radar occlusion state estimation with an accuracy of over 70% in 10 s and over 90% in 100 s.