Posez n’importe quelle question sur les cours, conférences, exercices, recherches, actualités, etc. de l’EPFL ou essayez les exemples de questions ci-dessous.
AVERTISSEMENT : Le chatbot Graph n'est pas programmé pour fournir des réponses explicites ou catégoriques à vos questions. Il transforme plutôt vos questions en demandes API qui sont distribuées aux différents services informatiques officiellement administrés par l'EPFL. Son but est uniquement de collecter et de recommander des références pertinentes à des contenus que vous pouvez explorer pour vous aider à répondre à vos questions.
Enhancing thermal comfort while reducing cooling demand is a pressing challenge for Mediterranean school buildings, especially as climate change accelerates the frequency and intensity of heat-wave events. Hybrid ventilation (HV), switching between natural ventilation (NV) and air conditioning (AC), can meet this challenge, but conventional passive or rule-based controllers cannot adapt to the shrinking potential of comfortable NV hours. This study proposes a reinforcement learning (RL) framework that continuously learns when to deploy NV and when to shift to AC set-points, thereby preserving comfort and efficiency under both present and projected future climates. A single-zone EnergyPlus model of a Beirut classroom is coupled to an RL environment with a 20 min control interval. The RL agent observes indoor temperature, predicted mean vote (PMV), and weather conditions. Its actions include (i) window opening and (ii) AC activation at different temperature setpoints. The reward function penalizes PMV deviation and HVAC energy consumption, but awards bonuses when comfort is achieved via NV. The agent is trained on a typical meteorological year (TMY) and evaluated on 2050 and 2080 hot-scenario weather files to test climate resilience. Ongoing simulations will benchmark the learned policy against a calibrated rule-based baseline, focusing on three metrics: percentage of time within the comfort band (−0.5 ≤ PMV ≤ +0.5), cooling energy demand, and indoor overheating. The study expects the RL approach to reveal adaptive control patterns, such as dynamically adjusting AC set-points in response to outdoor heat severity, and to demonstrate superior performance across current and future climates. By delivering a data-driven, climate-adaptive HV controller and an EnergyPlus testbed, this work aims to supply building operators and researchers with a practical pathway toward the resilient, low-energy operation of educational spaces in warming Mediterranean regions.