eduzhai > Applied Sciences > Engineering >

A Relearning Approach to Reinforcement Learning for Control of Smart Buildings

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 14 pages

Abstract: This paper demonstrates that continual relearning of control policies usingincremental deep reinforcement learning (RL) can improve policy learning fornon-stationary processes. We demonstrate this approach for a data-driven smartbuilding environment that we use as a test-bed for developing HVAC controllersfor reducing energy consumption of large buildings on our university campus.The non-stationarity in building operations and weather patterns makes itimperative to develop control strategies that are adaptive to changingconditions. On-policy RL algorithms, such as Proximal Policy Optimization (PPO)represent an approach for addressing this non-stationarity, but exploration onthe actual system is not an option for safety-critical systems. As analternative, we develop an incremental RL technique that simultaneously reducesbuilding energy consumption without sacrificing overall comfort. We compare theperformance of our incremental RL controller to that of a static RL controllerthat does not implement the relearning function. The performance of the staticcontroller diminishes significantly over time, but the relearning controlleradjusts to changing conditions while ensuring comfort and optimal energyperformance.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...