Real-time energy management for integrated energy systems with dynamic perception and correction of multi-dimensional uncertainties

Integrated energy systems constitute a reliable pathway for efficiently incorporating a variety of flexible resources under heterogeneous energy carriers while facilitating the integration and accommodation of renewable distributed generations. However, the multi-dimensional uncertainties originating from renewable distributed generations and energy demand impose considerable challenges on the secure and economic operation of integrated energy systems. To this end, this paper proposes a real-time energy management strategy for integrated energy systems by dynamically perceiving and correcting multi-dimensional uncertainties across different time scales. Specifically, a nominal model predictive control model is constructed to dynamically formulate operation reference trajectories wherein an integrated energy system operation scenario forecasting method is developed based on Wasserstein generative adversarial networks to realize dynamic perception of uncertainties. Subsequently, a cascaded ancillary model predictive control model is formulated from reference trajectories, enabling real-time correction of uncertainty disturbances most economically at a shorter time scale. Extensive numerical simulations and result discussions demonstrate the effectiveness and superior performance of the proposed solution.