Deep learning-based electric vehicle transmission system speed ratio prediction research
Lian-ge He,Jun Song,Yan Zhang,Xin-yang Wu
DOI: https://doi.org/10.1007/s11227-024-06748-9
IF: 3.3
2024-12-06
The Journal of Supercomputing
Abstract:This study is concerned with the analysis and research of powertrain parameter matching issues, as well as the selection design, of an SUV electric vehicle equipped with a two-speed transmission. Furthermore, the study encompasses a kinematics analysis of the entire vehicle, the establishment of the vehicle's driving mechanics model, and the derivation of the calculation method for key performance indicators of the vehicle. This paper employs the vehicle design requirements and conventional parameter matching process to ascertain the optimal drive motor selection. Furthermore, the paper presents a design methodology for crucial parameters, including drive motor power, motor torque, motor speed, and transmission speed ratio. Subsequently, the AVL-Cruise vehicle simulation software is employed to validate whether the selected drive motor and drive train speed ratio satisfy the vehicle's performance requirements. This is achieved by conducting simulation analyses under a range of conditions, including NEDC cycle conditions, acceleration time from 0 to 50 km/h, maximum speed, and maximum uphill climbing capacity. Subsequently, the veracity and viability of the simulations are validated. Furthermore, the main gear ratio, first gear speed ratio, and second gear speed ratio are validated through the use of the Cruise simulation software. Finally, 5179 sets of simulation results were obtained by matrix calculation of the main gearbox speed ratio, first gear speed ratio and second gear speed ratio with Cruise simulation software. The findings of this research indicate that by comparing vehicle simulation data samples with prediction results, the deep learning method can provide a novel design concept and methodology for R&D engineers during the forward development of vehicles.
computer science, theory & methods,engineering, electrical & electronic, hardware & architecture