Risk-Averse Generation Maintenance Scheduling for Power Systems Based on Contagious Value-at-Risk

Accurate quantification of uncertainty risks is pivotal for enhancing the reliability of generation maintenance scheduling (GMS) in power systems. However, existing risk quantification methods predominantly focus on the aggregate impact of uncertainty on system-wide operational risks, failing to identify critical risk sources. This limitation hinders the secure and efficient operation of power systems with high penetration of renewable energy. To address this issue, we propose a risk-averse GMS approach for power systems based on contagious value-at-risk (CoVaR). Specifically, we first introduce the CoVaR theory to identify dominant risk sources affecting the secure operation of the system and derive a general analytical expression for CoVaR that incorporates integral terms of uncertain variables. Subsequently, a scenario-based linearization reconstruction strategy is developed to discretize these integral terms, and the complex CoVaR model is reformulated into a computationally tractable mixed-integer linear programming (MILP) model. On this basis, a new risk-averse GMS model embedded with CoVaR constraints is constructed. This model achieves precise identification of critical risk sources by quantifying and comparing the impacts of different risk sources on both system operational costs and risk costs. Finally, simulation results on the modified IEEE 24-bus power system and IEEE 118-bus power system demonstrate the effectiveness and superiority of the proposed approach.