Short-Term Forecasting of Electricity and Gas Demand in Multi-Energy System Based on RBF-NN Model

To alleviate environmental pressures and improve energy efficiency, multi-energy system has become an important way of energy utilization. Accurate energy loads forecasting has significant effect on the economic scheduling and optimal operation of multi-energy system. Thus, this paper proposes a short-term forecasting method of electricity and gas demand in multi-energy system based on Radial Basis Function Neural Network (RBF-NN) model. Firstly, the correlation analysis of power and gas loads and electricity price characteristics of the multi-energy system is carried out. Then, the power load, gas load and electricity price time series are established, fully considering the effects of the coupling relationship between any two kinds of loads, together with electricity price, weather and other factors. After that, the network model structure of RBF-NN is introduced. K-means clustering algorithm is used to determine the appropriate data center for the radial basis function of hidden layer nodes. Finally, the model is verified via the practical data of a multi-energy system in a certain park in China. Comparing with the univariate prediction method, it is verified that the multi-energy system load prediction method proposed in this paper can effectively consider the coupling relationship between power and gas loads, and improve the prediction accuracy.