Reinforcement Learning Algorithm Outperforms Traditional Methods in Active Noise Control Systems

Strategic Insights -

Researchers from Shenyang Ligong University have developed a reinforcement learning (RL) algorithm for secondary path identification in active noise control systems, which outperforms traditional methods in reducing noise levels. The study, published in AIP Advances, found that the RL algorithm significantly improved noise reduction by 6.8 dB and reduced the average error by 37.2%. The research has significant implications for the development of emerging technologies, particularly in the field of machine learning.

Key Takeaways:

  • The researchers proposed a secondary path identification method based on reinforcement learning (RL) and periodic prediction mechanisms to address the issue of inaccurate secondary path modeling in traditional active noise control systems.
  • The RL algorithm was found to outperform traditional algorithms in reducing noise levels, with a significant improvement of 6.8 dB in overall noise reduction.
  • The study concluded that the RL algorithm reduced the average error by 37.2%, increased the signal-to-noise ratio by 5.1 dB, and reduced the relative error by 12.6%.
  • The research was conducted by a team of researchers from Shenyang Ligong University, led by Wenjun Li, and included Caiyun Wu and Fan Bai.
  • The study was published in AIP Advances, a journal published by AIP Publishing LLC.
  • The research has significant implications for the development of emerging technologies, particularly in the field of machine learning.

Statistics:

  • The RL algorithm improved overall noise reduction by 6.8 dB.
  • The average error was reduced by 37.2%.
  • The signal-to-noise ratio was increased by 5.1 dB.
  • The relative error was reduced by 12.6%.

Sources:

  • NewsRx. Research from Shenyang Ligong University Provides New Data on Algorithms (Reinforcement learning algorithm for secondary path identification in active noise control systems). Journal of Engineering. October 20, 2025; p 3122.
  • AIP Advances. Reinforcement learning algorithm for secondary path identification in active noise control systems. 2025,15(8):085021-085021-11.