With the intermittent and unstable renewable power feeding into the district energy systems (DES), the reliability of the system need to be accurately evaluated is of great significance. In order to predict the probability of system operational state in the design stage, in addition to provide the reasonable distributions of the input parameters, the transmission of the uncertainty in the analytical model need to clarify. In this paper, taking photovoltaic systems for example, the method to quantify the meteorological parameters distributions in the uncertainty analysis was proposed. And then the transmission of the uncertainty in the theoretical model and data-driven model were compared. The Back Propagation Neural network model (BP model) was selected as example. The BP model shows high accuracy than theoretical model, meanwhile, it also shows lower uncertainty. The results indicated that the data-driven model is more suitable for estimating the system output in the design stage. The research will provide guidance for system modeling by using data-driven model.
Keywords district energy systems, uncertainty transmission, theoretical model, BP model, photovoltaic