International Journal of Automotive Technology最新文献

The gearbox of an electric vehicle operates under the high load torque and axial load of electric vehicles. In particular, the bearings that support the shaft of the gearbox are subjected to several tons of axial load, and as the mileage increases, fault occurs on bearing rolling elements frequently. Such bearing fault has a serious impact on driving comfort and vehicle safety, however, bearing faults are diagnosed by human experts nowadays, and algorithm-based electric vehicle bearing fault diagnosis has not been implemented. Therefore, in this paper, a deep learning-based bearing vibration signal processing method to diagnose bearing fault in electric vehicle gearboxes is proposed. The proposed method consists of a deep neural network learning stage and an application stage of the pre-trained neural network. In the deep neural network learning stage, supervised learning is carried out based on two acceleration sensors. In the neural network application stage, signal processing of a single accelerometer signal is performed through a pre-trained neural network. In conclusion, the pre-trained neural network makes bearing fault signals stand out and can utilize these signals to extract frequency characteristics of bearing fault.

Safety, comfort, range, and energy consumption continue to be highly important for today’s motor vehicles. This study considers suspension systems and investigates a semi-active suspension system based on a tubular flux-switching linear machine. The study optimizes motor performance by defining objective functions that use genetic algorithms to reduce cogging forces. Different configurations of model including block, circular, and cylindrical magnetized have been compared in terms of flux density, mesh size, and manufacturing cost by using magnetostatic analyses. Changes in induced voltage, cogging force, and thrust force according to the current based on 2D transient time analysis data were investigated. Multi-physics analysis of the machine was performed on a quarter-vehicle model using linear analysis, as using a linear machine was more effective for vibration mitigation. A prototype of the proposed block magnet-configurated machine was manufactured, comprising a linear motion system driven by an induction motor with a crankshaft to simulate linear motion of the suspension system. Analysis shows that the designed machine is effective as a semi-active suspension system for vibration and damping.

To improve emissions inventory, design a real-work operation mode that simulates the operating characteristics of an excavator. The test was conducted at a specialized construction machinery test site to maintain constant operator and soil conditions. Engine speed and actual engine percentage torque were obtained from the onboard diagnostic terminal through a data acquisition device. In Korea, a fixed LF of 0.48 is uniformly applied to all construction machinery. However, it may not be entirely reasonable to use this LF for construction machines performing a variety of tasks. In practical operation tests conducted on two excavators, the LF was measured as 0.426 and 0.47, demonstrating that the fixed LF may not always be applicable. By implementing an LF that is subdivided for each specific type of construction machine, the error in emission calculations could potentially be reduced by 2% to 12%.

This paper presents a joint sliding mode control algorithm with fuzzy adaptive gain to address the problem that the lateral stability of distributed drive electric vehicles is affected by system parameter perturbation and external environment disturbances under steering conditions. The control system is designed by considering the influence of road conditions and tire nonlinearity, taking the yaw rate and sideslip angle as control variables. The difference between the expected value and the actual value of the control quantity is taken as the input to obtain the expected front-wheel angle for feedback correction. Aiming at the problem that it is difficult to obtain the critical driving state parameters of vehicles and to directly measure the road adhesion coefficient which affects the vehicle's lateral stability, this paper presents a simplified unscented Kalman filter observer which is designed to dynamically estimate the vehicle state parameters and road adhesion coefficient for the lateral stability controller. Based on CarSim and MATLAB/Simulink, a co-simulation model is developed and verified under different working conditions. The results reveal that the proposed lateral stability control algorithm effectively reduces the front wheel steering angle, improving the vehicle's handling stability while reducing the driver's operating burden and improving driving safety.

A method based on strong tracking unscented Kalman filter with multiple suboptimal fading factors (MSTUKF) was proposed to accurately estimate the state of charge (SOC) of power batteries of electric vehicles online. Taking a certain lithium-ion battery as the research object, a second-order RC equivalent circuit model of the battery was established based on its external characteristics and related mechanism. Then the recursive least squares method with forgetting factor was adopted to identify the model parameters, and the MSTUKF nonlinear state space equation of the battery was established according to the equivalent circuit model. Finally, the SOC estimation algorithm was verified by simulation experiments under ECE15 and UDDS conditions. The results show that the error of MSTUKF in SOC estimation of lithium-ion battery is kept within 1.5%, so this method can estimate battery SOC accurately.

To analyze the car-following behavior accurately, this paper takes the drivers’ psychological factors into the consideration based on the psychological field theory. The vehicle dynamics indexes are extracted through vehicle history trajectories and the driving styles are clustered by k-means methods. After that the perception coefficient, reaction coefficient, and driving style correction coefficient are obtained and integrated into the psychological field for characterizing the drivers’ driving styles. The psychological car-following model which considers the driving styles is built based on the Full Velocity Difference (FVD) model. Finally, the model is validated under the MATLAB/Simulink environment and the result reveals that the psychological field car-following model achieves higher accuracy of characterizing car-following behavior compared with the FVD model and the interaction potential model.

A combined steering and braking collision avoidance control method based on model predictive control is proposed in the present paper. This method can overcome the limitations of single steering or braking collision avoidance and improve vehicle safety and tracking performance in extremely complex conditions. The utilization rate of tire force, path tracking performance, and driving stability are considered in the design of the model predictive control (MPC) optimization objective. In addition, the front wheel angle and four-wheel braking torque are taken as the optimization variables. Simulation results showed that the proposed method can greatly improve the path tracking performance and driving stability of vehicles.

With the rapid advancement of AI technology, speech recognition has also advanced quickly. In recent years, speech-related technologies have been widely implemented in the automotive industry. However, in-vehicle environment noise inhibits the recognition rate, resulting in poor speech recognition performance. Numerous speech enhancement methods have been proposed to mitigate this performance degradation. Filter-based methodologies have been used to remove existing vehicle environment noise; however, they remove only limited noise. In addition, there is the constraint that there are limits to the size of models that can be mounted inside a vehicle. Therefore, making the model lighter while increasing speech quality in a vehicle environment is an essential factor. This study proposes a Wave-U-Net with a depthwise-separable convolution to overcome these limitations. We built various convolutional blocks using the Wave-U-Net model as a baseline to analyze the results, and we designed the network by adding squeeze-and-excitation network to improve performance without significantly increasing the parameters. The experimental results show how much noise is lost through spectrogram visualization, and that the proposed model improves performance in eliminating noise compared with conventional methods.

To solve the shortcomings of the existing hub motors in the practical application of electric vehicles, an integrated dual-rotor hub motor (DRHM) was proposed, which can realize multiple drive modes to adapt to the vehicle's variable driving conditions. Aiming at the complex structure of the DRHM, a multi-objective optimization method of design variables stratification based on comprehensive sensitivity was proposed. The design variables with medium and high sensitivity were optimized by the response surface method and genetic algorithm, respectively. After overall weighing the optimization objectives of output torque, torque ripple, usage amount of permanent magnets and magnetic coupling coefficient, three candidate design were screened out. By the comprehensive performance evaluation of the motor, the optimal structural sizes were determined. Based on a two-dimensional model, the electromagnetic performances of the DRHM were analyzed. The simulation results show that the motor has a small cogging torque and low magnetic coupling degree, and the independent control and stable operation of the internal and external motors can be realized. Besides, the basic characteristics of the DRHM prototype were tested. The experimental results accords well with the simulation results, which show that the proposed motor structure is reasonable and the multi-objective optimization method is effective.

In the developmental phase of passenger automobile hoods, it is crucial to take dent resistance and pedestrian head protection performances into account. And maintaining a performance balance between the two aspects has proven challenging. With few studies on how to effectively maintain the balance, in this paper, a certain passenger car hood was used as a basic model to investigate structural improvement directions that benefit both performances, by modifying several key variables of the honeycomb shaped inner panel one at a time and then outputting results of hood dent resistance and headform impact through Abaqus and LS-DYNA, respectively. The results indicated that raising the honeycomb inner panel structure at positions with poor stiffness contributes to improvements in dent resistance and pedestrian head protection performance, reflecting in a maximum stiffness increase of 50% and a reduction of 2.98% in the average value of HIC₁₅ change rate. And other alternative improvement options, as well as their effects on dent resistance and pedestrian head protection performance, were provided, providing insights for optimizing the structure of vehicle hood inner panels.