Identification of Critical Security Boundary for Resilient Power Systems Driven by Model-Data Fusion

Abstract

With the evolution of power system structure and the expansion of renewable energy scale, the security and stability challenges brought by extreme events are becoming increasingly prominent. The traditional transient model of power systems is tailored specifically for certain fault scenarios and exhibits nonlinear characteristics. Consequently, its solutions are often characterized by a time-intensive nature and suboptimal generalization performance. Therefore, a security boundary identification method for resilient power system driven by model-data fusion is proposed in this paper. Based on the security constrained unit commitment model of power system, the umbrella constraint identification method is employed to identify the effective constraints. A massive extreme sample set based on the dynamic response model of the CloudPSS platform is established, and support vector machine is leveraged to identify and extract transient safety constraints. The critical security boundary is characterised by the combination of umbrella constraints and transient safety constraints, which can be embedded into the economic dispatch model to facilitate the secure and efficient operation of power system. Case studies based on IEEE-39 systems verified the effectiveness of the proposed method in different fault scenarios.