To address the issue of discontinuous curvature in autonomous parking path planning, this paper analyzes vehicle kinematics, and combines the arc-line-arc planning method with the reverse parking process. A fifth-degree polynomial optimization approach is employed to generate a compact parking trajectory with continuous curvature. To enhance parking tracking accuracy, the discrete LQR tracking controller based on the kinematic model is improved using fuzzy control methods. Simulations and experimental validations are conducted to verify the effectiveness of the algorithm. In the Simulink/CarSim co-simulation, the maximum tracking error is 0.027 m, and the average tracking error is 0.013 m. In real-vehicle experiments, the maximum tracking error is 0.07 m, and the average tracking error is 0.029 m. Compared to the LQR tracking controller, the FUZZY-LQR tracking controller reduces the average tracking error by 33%, improving the autonomous parking path tracking performance.

A multi-objective optimized automatic design process is developed based on the airflow velocity required for cooling performance in the passenger cabin. This process considers the airflow performance on the driver's side and the passenger's side, focusing on the positioning of the grille vent blades. Based on CFD simulations and a multi-disciplinary optimization design platform, the Latin hypercube sampling method is used to generate sample points and construct the DOE matrix. A neural network-based proxy model is then built to predict the blow-face airflow velocity performance parameters. The NSGA-Ⅲ algorithm is used to obtain the Pareto frontier diagram for the multi-objective optimization problem. The optimized grille blade position increases the airflow speed by 109.1% on the driver's side and by 137.5% on the front passenger's side. The reliability of the optimization results is verified through unsteady CFD simulations and cooling performance tests before and after the optimization.

Taking a domestic fuel commercial vehicle as an example, an energy consumption optimization prediction model suitable for commercial vehicles was constructed using the Internet of Vehicles big data platform and a neural network model. Firstly, the historical vehicle operation data was preprocessed to analyze the correlation between different vehicle operation characteristic data. Secondly, an adaptive weight attention mechanism was introduced based on Bi-directional Long Short-Term Memory (BiLSTM) and the characteristics of vehicle data. The Improved Whale Optimization Algorithm (IWOA) was used to optimize the network hyperparameters of the model, leading to the construction of the IWOA-BilSTM-Attention commercial vehicle energy consumption optimization prediction model. Finally, the prediction performance of multiple models under different driving conditions were compared and analyzed. The results show that under actual driving conditions, the root mean square error and the mean absolute error of the optimized model are reduced by approximately 26.73% and 20.0%, respectively, compared with the original model. This verifies the feasibility of the optimized model for predicting the energy consumption of commercial vehicles.

Neural networks lack interpretability and the D-S theory is prone to paradoxes in high-conflict scenarios of multimodal fusion. In response, this paper proposes a result-level multimodal fusion method that integrates a confidence estimation network with an improved D-S theory. The method consist of two key components. First, a confidence estimation network reframes the classification problem in target detection as a confidence estimation task, providing confidence scores for the detection results of individual unimodal networks. Second, a fusion method with improved D-S theory uses confidence scores and class information to construct evidence, achieving final fusion of detection data from different modalities. Evaluation experiments on the KITTI dataset show that the proposed fusion method improves mAP by up to 6.64% compared to image-based detection and up to 15.43% compared to point cloud-based detection. In the comparison of fusion methods, the proposed fusion method achieves an mAP improvement 0.81% higher than the classical D-S fusion. It effectively reduces classification conflicts and addresses the limitations of the classical D-S theory.

Aiming at the instability of high-speed vehicles in windy and rainy environments, the Euler-Lagrange method was used to numerically simulate the external flow field of automobiles under such conditions. The aerodynamic characteristics was investigated at different side wind speeds and rainfall intensities. The whole vehicle dynamics model was established and subjected to aerodynamic loads. The lateral displacement of the vehicle under the windy and rainy conditions was calculated. The results show that, at a constant side wind speed, rainfall increases the drag vortex and the wake vortex diffusion region on the leeward side. Changes in the vortex structure increase the negative pressure area on the leeward side of the body and reduce the the trailing airflow velocity. As a result, aerodynamic drag and side forces are increased. Additionally, changes in road surface adhesion further reduces the vehicle's lateral stability.

To address the issues of poor yaw stability and low trajectory tracking accuracy in in-wheel motor-driven vehicles under complex operating conditions, a coordinated control method was proposed. By using neural networks for dynamic identification, the stable region in the phase plane of vehicle center-of-mass sideslip angle and sideslip angular velocity was determined. And the instability factor was obtained based on the boundary line features. This factor served as a parameter influencing the objective function weight, while sliding mode control was employed to prevent excessive wheel slip. Simulation results show that compared to the single-target MPC trajectory tracking control strategy, the proposed method reduces the maximum lateral tracking error on low-adhesion road surfaces by 61.7%, and decreases the maximum sideslip angle of the vehicle's center of mass by 75.7%. Even at high speeds, the vehicle maintains stable motion, achieving a balanced improvement in both trajectory tracking accuracy and yaw stability.

Rack force is a critical parameter in road feel design and steering follow-up control for steer-by-wire systems. Since it is difficult to measure rack force directly in mass-produced vehicles, its estimation becomes a key aspect in steer-by-wire. This article adopts two methods to estimate the rack force. The first method is based on vehicle dynamics, using Luenberger observer to obtain lateral velocity and the tire brush model is applied to analyze steering resistance torque. The second method is based on steering dynamics, involving a dynamic model for the steering actuator and the direct estimation of rack force using a Kalman filter. The results of the two methods are compared through hardware-in-the-loop (HIL) testing, and then a rule-based fusion strategy using multiple variables is proposed to combine the strengths of the two methods. Finally, bench test results show that the proposed fusion strategy effectively improves the accuracy and real-time performance of rack force estimation under different working conditions.

The rapid development of new energy vehicles has dramatically increased the demand for battery materials such as lithium, cobalt, nickel, and manganese. To assess the supply risks of these resources, the demand for battery materials in China's new energy passenger vehicles from 2023 to 2050 was predicted using the Gompertz curve model and material flow analysis, under two different scenarios of battery technology development. The results indicate that by 2050, under the lithium iron phosphate route (LFPR), the demand for lithium, cobalt, nickel, and manganese will be 238, 169, 362, 158 kilotons, respectively. Under the nickel-manganese-cobalt (NMC) lithium battery route, the estimated demands will reach 242, 201, 1 084, 187 kilotons, respectively. Recycled lithium, cobalt and nickel in 2050 are expected to meet at least 86.5%, 93.5% and 65.8% of their annual demand, respectively. Given the current lack of comprehensive laws and regulations for waste battery recycling in China, it is essential to develop relevant standards.

To address the difficulty in determining the optimal return angular speed at various vehicle speeds in traditional active Return-to-Center (RTC) control, a new active RTC control method for steer-by-wire systems is proposed, combining a return speed reference model and sliding mode control. A time-window-based mechanism for determining the active return-to-center state is designed. The return speed reference model is established based on the tire aligning torque, and an active RTC sliding mode control strategy is developed accordingly. Hardware-in-the-loop simulation results indicate that the proposed active return state determination mechanism can accurately switch system states, the return speed reference model exhibits high accuracy, and the sliding mode control strategy effectively ensures that the steering wheel reliably follows the reference return speed.

Parking tracking accuracy directly affects parking safety, efficiency, and available parking space. Currently, most autonomous parking path tracking relies on model-based feedback control. High tracking errors can arise from a decline in the algorithm's control performance due to uncertainties in system model parameters. In this paper, a feedforward control approach based on iterative learning was developed to reduce the impact of model parameter uncertainty on parking path tracking. Considering that iterative learning control of the system in the time domain was usually affected by the actual speed of the actuator, the system was transformed from the time domain to the space domain, which was related to the desired path. Due to the difficulty in measuring some state variables in the system model and the system's failure to meet the D-type iterative learning rate convergence condition, the design criteria for an H_∞ observer were proposed to accurately estimate state information. Meanwhile, an augmented system with observation errors was constructed to implement iterative learning control, which further reduced the parking path tracking error based on the initial parking tracking information from linear quadratic optimal control (LQR). Finally, a hardware-in-the-loop (HIL) test was established, which proved that the proposed method had excellent practical application potential. The experimental results show that after several iterations, the proposed control method tracks the desired path more accurately than the initial LQR control.