Simulation and Prediction of Thermal Energy Demand

摘要

This paper presents the development and exploitation of two mathematical models based on statistical methods and artificial neural networks for analyzing and predicting the thermal power of buildings connected to a substation supplied by a district heating system. Both models are able to accurately capture the non-liner dynamics of thermal power demand that depends on very different technical and subjective human factors. The two models are compared taking into account the correlation coefficient R as a performance criteria and are validated by evaluating the accuracy of the approximation for other experimental data than those used in modeling stage. Predicted and experimental values for each model are well matched and highlight the success of applying statistic and neural networks models in predicting thermal power demand of buildings.