Intelligent Air Cooling Control in Thermal HVAC Systems Using Deep Learning

Occupancy-based control of Heating, Ventilation, and Air Conditioning (HVAC) systems with the use of machine learning is described. The common HVAC systems tend to be inefficient because they are either programmed or sensor-driven and thus they waste energy and poor the indoor environmental setup. With real-time occupancy data, HVAC systems can also adjust on a real-time basis by adjusting heating, cooling, and ventilating based on the presence or absence of occupants. This paper suggests a machine learning-based model using the UCI Energy Efficiency Dataset to model and predict performance with regard to cooling. Several models were assessed and compared based on standard regression measures, such as MLP, NN, LSSVR, and Random Forest (RF). The Random Forest model provided the highest level of accuracy and strength with an R² of 96.4 and very low prediction errors (MAE = 0.01, MSE = 0.01, RMSE = 0.076). As the outcomes show, the suggested RF-based model successfully determines the nonlinear dependencies between thermal and environmental factors, which is why it can be deemed as a sound solution to air cooling control and optimization of air conditioning systems with consideration of energy efficiency.