Auxiliary Brake Apparatus by Air Compression and Release for Stop of Heavy FCV Regenerative Brake (6th report)
-Proposal of a Rotary Vane Mechanism for Compact Design-

Ko Amada･Daiki Fujita･Mimi Tsuchiya･Toshinori Fujita･Takashi Shibayama (Tokyo Denki University)

In our previous reports, we proposed a simplified auxiliary brake based on an air compression and release system to be used in situations where regenerative braking in FCV trucks is unavailable. The previous design employed an axial piston mechanism, which required converting rotational motion into reciprocating motion, thereby imposing limitations on compactness. In this report, we propose a rotary vane mechanism for improved compactness and present its characteristics.

Development of a Power Transmission Characteristics Testing Machine for High-Speed, High-Output, and High-Reduction-Ratio Traction Drives

Natsuki Morita･Takeshi Yamamoto (Tokai University)

To evaluate the performance of traction drives, a power-circulation testing machine was developed to enable measurements at 50,000 rpm and 50 kW. This machine overcomes facility constraints and achieves traction coefficient measurements under these specific high-speed and high-output conditions. Furthermore, in tests under high reduction ratios, it was confirmed that the traction coefficient could be successfully measured without any reduction.

Electric Control of a CVT for Motorcycles Using a Self-Locking Reduction Mechanism

Genjiro Ueda･Tatsuhito Aihara (Hosei University)

An electric control method for a CVT for motorcycles using a self-locking reduction mechanism and a motor is proposed. The pulley position is controlled by the motor to achieve rapid shifting and high efficiency. In addition, the self-locking function enables the pulley position to be maintained without power consumption. This presentation reports the mechanism configuration, operating principle, and shifting characteristics of the proposed system.

Powertrain Vibration Suppression via the Integration of Domain Randomization-Based Deep Reinforcement Learning Control and Model-Based Control

Heisei Yonezawa･Ansei Yonezawa･Itsuro Kajiwara (Hokkaido University)

The design of control systems for powertrains requires consideration of nonlinearities and uncertainties in dynamic characteristics. In recent years, deep reinforcement learning has attracted considerable attention as a promising approach; however, learning inefficiency and policy conservativeness remain significant challenges. To address these issues, this study proposes a method for assisting domain randomization-based deep reinforcement learning by introducing a model-based controller that ensures baseline performance.

Development of a Mode-Adaptive Shift Control Optimization Method for Fuel Cell Garbage Trucks with Multi-Layered Operating Modes

Yida Bao･Jiakai Gao･Xiang Zhang･Jiachen Xi･Yiyuan Fang･Wei-Hsiang Yang･Yushi Kamiya (Waseda University)

Fuel cell garbage trucks exhibit complex collection-and-driving patterns with wide-ranging performance requirements, making it difficult for a single shift map to maintain optimal operation. This study proposes a multi-layered operation-mode-based shift map considering motor torque limit characteristics and regenerative braking. Bayesian Optimization is applied to jointly optimizes gear ratios and shift maps under an integrated objective combining energy efficiency, dynamic performance, and drivability. Simulation results on representative duty cycles show that, compared with conventional fixed shift maps, the proposed approach improves both energy efficiency and dynamic performance while preserving drivability.

Proposal of a CVT Shift Map for BEVs Based on Optimal Control Methods

Hanqing Zhao･Yasuo Moriyoshi･Tatsuya Kuboyama (Chiba University)

For EV CVT control, shift maps are generally designed empirically, and their optimality has not been sufficiently investigated. In this study, optimal CVT ratios under various driving conditions were calculated using an optimization method, and the differences from the conventional CVT map were analyzed. In addition, correction regions were classified using a decision tree, and correction maps for each region were constructed to investigate a method for improving conventional CVT shift maps.

Development of a Virtual Performance Evaluation Model for New Multi-Mode Hybrid Powertrain System with RCP-HIL Validation

Myeong-Won Seo･Dohyun Park･Janghyeok Won･Beomho Lee･Taeho Park･Deokjin Kim (Korea Automotive Technology Institute)

This study presents a virtual performance evaluation model for a multi-mode hybrid powertrain comprising a 1.6L GDI engine, P1/P3 electric motors, and a dog-clutch 2-speed transmission. Three driving modes - EV, Series HEV, and Parallel HEV - are supported through a multi-controller architecture (HCU, TCU, ECU, MCU, BMS, BCU). Plant and controller subsystems are separated into non-virtual blocks, enabling RCP-HIL deployment via dual Speedgoat machines over CAN bus. All performance targets were met: (FTP-75, HWFET), 0-100 km/h in 10.00 s, EV grade ability 39.14%, and maximum speed 168.7 km/h.