Neutral-Connect Control in a Two-Speed Transmission Based on Demand Torque Prediction Using a Time Series Deep Learning Model

• Published in: International journal of automotive technology 2025-02, Vol.26 (1), p.47-61
• Main Authors: Ahn, Jihyeok, Gwak, Seoku, Jeong, Seyoung, Kim, Kyung-Ho, Hwang, Sung-Ho
• Format: Article
• Language: English
• Subjects: Acceleration; Algorithms; Automotive Engineering; Connected Automated Vehicles and ITS; Deep learning; Electric; Electric vehicles; Engineering; Fuel Cell; Hybrid Vehicle; Machine learning; Powertrain; Regenerative braking; Time series; Torque; Transmissions (automotive); Vehicle Dynamics and Control; Virtual environments
• ISSN: 1229-9138; 1976-3832
• DOI: 10.1007/s12239-024-00143-2

Many types of electric vehicle powertrains are being researched. Among them, devices that can transition to two-wheel drive (2WD) are being introduced because full-time all-wheel drive (AWD) is inefficient in power consumption. In addition, transmissions are being applied to reduce the required motor capacity and increase the performance of the vehicle. In this study, we propose a method to transition between AWD and 2WD using the neutral state of the transmission. Because the drive mode transition is based on the driver's demanded torque with the maximum torque of the front wheel motor, the drive mode transition can be delayed. To this end, we propose a method to predict the driver's demanded torque using a time series deep learning model and transition the drive mode in advance. The time series deep learning model is trained using data generated in a virtual environment that includes external sensor information. By comparing the predictive value-based algorithm using the learned model and the typical algorithm based on driver input, we found that the predictive value-based algorithm predicts the transition faster than the typical algorithm, improving acceleration response and regenerative braking energy.