WSN-Based Data-Driven Digital Twin for Energy Efficient HVAC Systems

Efficiently managing of energy consumption while ensuring a comfortable indoor environment is a crucial challenge in Heating Ventilation and Air Condition (HVAC) system control. Traditional approaches rely on physical models, while effective to some extent, entail laborious, time-consuming processes and lack scalability, imposing a substantial burden on HVAC system controllers. In this paper, we propose a data driven approach combined with Wireless Sensor Network (WSN) and Digital Twins (DT), allowing for the representation of system components in both the physical and virtual worlds. More specifically, a Long Short-Term Memory (LSTM) is integrated to enable in-advance modeling and prediction of indoor temperature for a potential Model Predictive Controller (MPC). The LSTM model is trained and validated using a dataset collected from sensors deployed in a candidate building. Several benchmarks were evaluated to ensure the efficiency of the proposed system model.