Scholarlink

This study proposes an adaptive PID control strategy for industrial processes by integrating Deep Reinforcement Learning (DRL) with conventional PID controllers, particularly focusing on wastewater treatment. By leveraging the strengths of DRL, specifically Proximal Policy Optimization (PPO) and Deep Deterministic Policy Gradient (DDPG), the proposed DRL-PI framework adaptively adjusts PID parameters to enhance control performance under nonlinear and time-varying conditions. Experiments using the BSM2 simulation model demonstrated that the DRL-PI framework outperformed traditional PI controllers in terms of stability, response accuracy, and adaptability. The results highlight the potential of DRL-based adaptive control in complex industrial environments.

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

Adaptive PID Controller for Industrial Process Based on Reinforcement Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Yicong Yang,
  • Qili Chen

摘要

This study proposes an adaptive PID control strategy for industrial processes by integrating Deep Reinforcement Learning (DRL) with conventional PID controllers, particularly focusing on wastewater treatment. By leveraging the strengths of DRL, specifically Proximal Policy Optimization (PPO) and Deep Deterministic Policy Gradient (DDPG), the proposed DRL-PI framework adaptively adjusts PID parameters to enhance control performance under nonlinear and time-varying conditions. Experiments using the BSM2 simulation model demonstrated that the DRL-PI framework outperformed traditional PI controllers in terms of stability, response accuracy, and adaptability. The results highlight the potential of DRL-based adaptive control in complex industrial environments.