Deep Reinforcement Learning-Based Optimization Strategy for Smart Building Operations

In recent years, with the increasing importance of building energy consumption and energy-saving management, building energy consumption and carbon emissions in China account for approximately 20% of the national total. The application of intelligent control systems to optimize building operational efficiency has gained attention. These systems improve building load demand response while reducing operational costs. This study designs an adjustable load control model for building operations based on deep reinforcement learning (DRL) technology, aiming to optimize both electricity cost and indoor temperature comfort. Using the Soft Actor-Critic (SAC) algorithm as a stable baseline model, the agent can intelligently adjust air conditioning loads under fluctuating electricity prices and temperature constraints to minimize building operating costs. To test the model’s effectiveness, simulations were conducted on a test dataset containing actual load data. The results indicate that the model can effectively maintain indoor temperature within the target range over multiple periods while significantly reducing overall building load. This simulation validates the potential of deep reinforcement learning in complex building energy management and offers new insights for the development of building energy-saving control systems.