Artificial neural network-based distributed optimal control for building HVAC systems

Heating, ventilation, and air conditioning (HVAC) systems are the major contributors to energy consumption in buildings, significantly impacting energy efficiency and occupants’ thermal comfort. In this context, this paper presents a data-driven distributed control framework that utilizes artificial neural networks (ANNs) to model the multi-time-step thermal dynamics of buildings. Each building has a local controller to minimize electricity generation costs in the microgrid while maintaining individual thermal comfort. To coordinate these local controllers, a distributed strategy is adopted to manage HVAC operations. Although conflicting objectives may arise during the iterative distributed coordination process, the framework effectively balances energy usage among the buildings and achieves convergence in total power consumption. Simulation results demonstrate that the ANN-based approach optimizes HVAC operations, achieving the aims of temperature regulation and optimal energy management.