Efficient multi-objective optimization of gear ratios and motor torque distribution for electric vehicles with two-motor and two-speed powertrain system

• Published in: Applied energy 2020-02, Vol.259, p.114190, Article 114190
• Main Authors: Kwon, Kihan, Seo, Minsik, Min, Seungjae
• Format: Article
• Language: English
• Subjects: Electric vehicles; Gear ratio; Multi-objective optimization; Surrogate model; Torque distribution; Two-motor and two-speed powertrain
• ISSN: 0306-2619; 1872-9118
• DOI: 10.1016/j.apenergy.2019.114190

•Developing an EV model including two-motor and two-speed powertrain system.•Effect analysis of gear ratios and torque distribution on the driving requirements.•Formulating an multi-objective optimization problem for efficiency and performance.•Constructing the surrogate models using an effective adaptive sampling method.•Obtaining the Pareto solutions including optimal gear ratios and torque distribution. In an electric vehicle (EV), a two-motor and two-speed powertrain system is superior to other powertrain systems in terms of the driving requirements, achieving an excellent dynamic performance and energy efficiency. Because the most important design specifications for a two-motor and two-speed powertrain are the motor torque distribution between the two motors, and the first and second gear ratios, these specifications should be optimized to improve both performance and efficiency as much as possible. To analyze such requirements, an EV model, including two-motor and two-speed powertrain system, was constructed. The acceleration time and energy consumption were employed as the evaluation criteria for the quantification of performance and efficiency, respectively, and the analysis results when changing the gear ratios and the torque distribution, show that these specifications significantly influence on the performance and efficiency. Therefore, an optimization of gear ratios and torque distribution is essential for achieving a superior powertrain system of an EV. Because of the trade-off relationship between the performance and efficiency, a multi-objective optimization problem is formulated to minimize the acceleration time and energy consumption. To decrease the excessive computational effort during a multi-objective optimization process, efficient surrogate models of each objective function were developed using an artificial neural network and an adaptive sampling method. The surrogate model-based optimization was performed, and the optimization results show a Pareto front that provides a variety of optimal solutions between the objective functions, as well as the validity of the surrogate model-based multi-objective optimization.