Scholarlink

To address the challenges posed by high temperatures, high pressures, and vibrations during aircraft operation, this paper proposes a model reference adaptive system (MRAS) control strategy based on deep reinforcement learning for sensorless vector control. This strategy, designed for permanent magnet synchronous motors (PMSM), utilizes a deep deterministic policy gradient (DDPG) reinforcement learning intelligent compensator based on the Actor-Critic architecture. It aims to enhance dynamic performance and reduce tracking errors caused by MRAS estimation errors and non-optimal control parameters. The proposed controller demonstrates increased robustness by incorporating real-time online learning and training with extensive system data, ensuring strict stability under various disturbances. Simulation results validating the effectiveness of the proposed control strategy.

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

DDPG-Integrated MRAS for Sensorless Control of PMSM in Aircraft Electric Propulsion Systems

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Han Wu,
  • Bingqiang Li,
  • Dongheng Wang,
  • Saleem Riaz

摘要

To address the challenges posed by high temperatures, high pressures, and vibrations during aircraft operation, this paper proposes a model reference adaptive system (MRAS) control strategy based on deep reinforcement learning for sensorless vector control. This strategy, designed for permanent magnet synchronous motors (PMSM), utilizes a deep deterministic policy gradient (DDPG) reinforcement learning intelligent compensator based on the Actor-Critic architecture. It aims to enhance dynamic performance and reduce tracking errors caused by MRAS estimation errors and non-optimal control parameters. The proposed controller demonstrates increased robustness by incorporating real-time online learning and training with extensive system data, ensuring strict stability under various disturbances. Simulation results validating the effectiveness of the proposed control strategy.