Research on Neural Network Based Real-Time Thermal Load Prediction

For optimization and predictive control of HVAC systems, a real-time thermal load prediction model based on neural networks was researched. The influential factors of thermal load were analysed. As basic inputs in determination of the load, meteorological parameters were forecasted first. Then, a neural network was used to predict the thermal load of building under arbitrary meteorological conditions. On studying the generalization abilities of neural networks, the neural model was trained with "early stopping" method. The predictive network was used to predict the cooling load of a building in Beijing. Simulation results show that the neural network can predict real-time thermal load accurately, and the model can be used in HVAC system control.