A Mechanism-Based Data-Driven Interval Energy Flow Calculation Method for Integrated Energy Systems via Affine Arithmetic-Based Optimization

The multiple energy flows (MEF) in an integrated energy system (IES) have distinct physical properties, complex interaction mechanisms, and abundant sources of uncertainties, posing higher challenges for dispatchers to perceive system states from the present to a specific future. To accurately and efficiently quantify the impact of uncertainties on MEF, this paper proposes an affine arithmetic-based interval energy flow (IEF) calculation method, realized by a mechanism-based data-driven methodology. Firstly, an improved affine arithmetic-based optimization model of the electrical network is proposed, which implements data-driven quadratic approximation to guarantee the tractability of the optimization model. Next, affine arithmetic-based optimization models of gas and heating networks are realized by combining mathematical derivation and machine learning assistance. Additionally, the spatial branch-and-bound (SBB) algorithm guarantees the global optimality of each optimization process, facilitating computations in both grid-connected and islanded operational modes. Finally, numerical results on two IES study cases demonstrate that the proposed method outperforms other candidates, exhibiting superiority in terms of both speed and accuracy.