Model Predictive Control Method for an Integrated Community Energy System Based on a GRU-Driven Dynamic Energy Hub

In integrated community energy systems (ICESs), the efficiency parameters of energy conversion devices are influenced not only by current operating conditions (e.g., device load rate, temperature, and atmospheric pressure) but also by the dynamic characteristics of historical operating conditions. Neglecting the nonlinear dynamic relationship between the efficiency parameters and time series operating conditions can notably reduce the prediction accuracy for the efficiency parameters, thereby disrupting the energy supply-demand balance and reducing the economic performance of the ICES. To address this issue, a model predictive control (MPC) method is proposed for an ICES based on a gated recurrent unit (GRU)-driven dynamic energy hub (G-DEH). First, a G-DEH model is developed to capture the time series correlations in the operating conditions of the ICES, thereby accurately modeling the nonlinear and time-varying relationships between the efficiency parameters and these conditions. On this basis, an optimal scheduling model for the ICES, focused on enhancing the economic performance, is developed via the G-DEH model. The model is subjected to iterative rolling calibration through the MPC strategy, dynamically adjusting the degree of matching between the operating conditions and efficiency parameters, thereby enhancing the adaptability of the scheduling model to complex and dynamic operating conditions. The simulation results from a case study demonstrate that the proposed method significantly outperforms traditional methods in terms of the prediction accuracy for the efficiency parameters and the economic performance of the ICES. This confirms the significant value of the method in reducing operating costs and enhancing dynamic adaptability.