Scholarlink

<p>This article investigates the high-efficiency decision-making problem for the electric drive system of mining trucks equipped with integrated four motors and dual AMT transmissions under specific mining conditions. It aims to optimize overall vehicle performance by dynamically selecting optimal gear combinations and allocating motor torque in real time. Moving beyond traditional decoupled optimization approaches, an integrated co-optimization framework for multi-gear decision-making and multi-motor torque distribution is proposed. Furthermore, an expert-informed deep deterministic policy gradient (EI-DDPG) algorithm is designed, which incorporates a gearshift penalty term within the reward function to effectively balance energy consumption and shift frequency. Finally, a representative mining driving cycle is constructed based on real vehicle operation data, and hardware-in-the-loop (HIL) experiments are conducted under this cycle to validate the proposed algorithm. The results demonstrate that, compared with the traditional rule-based control strategy and the DDPG control strategy without expert guidance, the EI-DDPG strategy achieves significant performance improvements in terms of reducing energy consumption and decreasing shift frequency.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Torque and shift decision for quadri-motor mining truck based on expert-informed DDPG

  • 期刊论文

  • 发表日期：

    2026年5月3日

查看全文
  • Yongchang Li,
  • Ying Xin,
  • Chenghao Fang,
  • Tonglie Wu,
  • Feng Wang

摘要

This article investigates the high-efficiency decision-making problem for the electric drive system of mining trucks equipped with integrated four motors and dual AMT transmissions under specific mining conditions. It aims to optimize overall vehicle performance by dynamically selecting optimal gear combinations and allocating motor torque in real time. Moving beyond traditional decoupled optimization approaches, an integrated co-optimization framework for multi-gear decision-making and multi-motor torque distribution is proposed. Furthermore, an expert-informed deep deterministic policy gradient (EI-DDPG) algorithm is designed, which incorporates a gearshift penalty term within the reward function to effectively balance energy consumption and shift frequency. Finally, a representative mining driving cycle is constructed based on real vehicle operation data, and hardware-in-the-loop (HIL) experiments are conducted under this cycle to validate the proposed algorithm. The results demonstrate that, compared with the traditional rule-based control strategy and the DDPG control strategy without expert guidance, the EI-DDPG strategy achieves significant performance improvements in terms of reducing energy consumption and decreasing shift frequency.