An experimental evaluation of Deep Reinforcement Learning algorithms for HVAC control

TL;DR

This paper critically evaluates various Deep Reinforcement Learning algorithms for HVAC control, focusing on comfort, energy efficiency, robustness, and adaptability, highlighting their potential and challenges in real-world scenarios.

Contribution

It provides the first standardized, reproducible comparison of state-of-the-art DRL algorithms for HVAC control using the Sinergym framework.

Findings

DRL algorithms like SAC and TD3 show promise in complex HVAC scenarios.

Challenges include issues with generalization and incremental learning.

The study emphasizes the need for standardization in DRL evaluation for HVAC.

Abstract

Heating, Ventilation, and Air Conditioning (HVAC) systems are a major driver of energy consumption in commercial and residential buildings. Recent studies have shown that Deep Reinforcement Learning (DRL) algorithms can outperform traditional reactive controllers. However, DRL-based solutions are generally designed for ad hoc setups and lack standardization for comparison. To fill this gap, this paper provides a critical and reproducible evaluation, in terms of comfort and energy consumption, of several state-of-the-art DRL algorithms for HVAC control. The study examines the controllers' robustness, adaptability, and trade-off between optimization goals by using the Sinergym framework. The results obtained confirm the potential of DRL algorithms, such as SAC and TD3, in complex scenarios and reveal several challenges related to generalization and incremental learning.