A bilevel interval-stochastic robust optimization for integrated energy systems considering multiple heterogeneous uncertainties

Uncertainties are ubiquitous in integrated energy systems (IESs). Inadequate considerations and inappropriate descriptions of multiple heterogeneous uncertainties occurring throughout the optimization process of energy source, conversion, storage, and demand can lead to suboptimal operations or even system failures. Moreover, IESs typically feature a multiagent structure, comprising a few upper-level system operators and numerous lower-level users. To address heterogeneous uncertainties across agents, we consider their intrinsic characteristics and data availability across various processes and propose a hybrid interval-stochastic robust optimization method built on a bilevel framework. Specifically, in the upper level from the supply side, prediction uncertainties of renewable energy sources and performance uncertainties in energy conversion and storage equipment are characterized using a Wasserstein-based ambiguity set and Gaussian mixture distributions, respectively. In the lower level, response uncertainties on the demand side are described using interval numbers. Furthermore, a noniterative single-layer equivalent method is developed for the tri-uncertainty model, reformulating the complex hybrid bilevel nonlinear optimization problem into a tractable single-layer mixed-integer linear programming problem. Comprehensive case studies demonstrate that the proposed method accurately and completely characterizes multiple heterogeneous uncertainties in the operation optimization process of IESs, promoting decision reliability and increasing actual operational profit.