面向 HVAC 控制的安全且数据高效的基于模型的强化学习系统

基于模型的强化学习（MBRL）已被广泛用于建筑中的加热、通风和空调（HVAC）控制。一个关键挑战是需要大量数据才能有效训练用于 modeling building dynamics 的神经网络。本文提出了 CLUE，一个用于建筑 HVAC 控制的 MBRL 系统。CLUE 通过集成高斯过程（GP）模型来 modeling building dynamics 并具备不确定性感知能力来优化 HVAC 操作。CLUE 使用 GP 预测状态转移 as Gaussian distributions，有效地捕捉 prediction uncertainty 并 enhance decision-making under sparse data conditions。Our approach employs a meta-kernel learning technique to efficiently set GP kernel hyperparameters using domain knowledge from diverse buildings。这大大减少了在 HVAC 应用中通常与 GP 模型相关的 data requirements。此外，CLUE 将这些 uncertainty estimates 整合到 Model Predictive Path Integral（MPPI）算法中，enable the selection of safe、energy-efficient control actions。这种 uncertainty-aware control strategy evaluates and selects action trajectories based on their predicted impact on energy consumption and human comfort， optimize operations even under uncertain conditions。Extensive simulations in a five-zone office building demonstrate that CLUE reduces the required training data from hundreds of days to just seven while maintaining robust control performance。It reduces comfort violations by an average of 12.07% compared to existing MBRL methods， without compromising on energy efficiency。