基于变换器驱动稳定性约束的高比例逆变型电源电力系统发电扩容

Poorly damped converter-driven oscillations are becoming increasingly prevalent as conventional synchronous generation (SG) is progressively replaced by inverter-based generation (IBG). Future power systems should strategically optimize the investment decisions of IBG and SG to prevent unstable dynamic interactions. However, different combinations of candidate generators would introduce varying oscillation modes, complicating the task of reserving sufficient oscillation damping for various critical modes during generation expansion. To address this challenge, this paper formulates a constraint-learning-based generation expansion planning framework. First, a cost-dominant sampling method is proposed to generate diverse generation mix scenarios, encompassing various on-off states and outputs of candidate generators. Next, stability constraints are learned for each generator and then grouped into an ensemble through the proposed physical-ensemble constraint learning method. These constraints are dynamically integrated into generation expansion and activated based on the generator's on-off status to maintain converter-driven stability across various generation mix scenarios. The feasibility and effectiveness of the proposed framework are validated on a modified IEEE standard test case and a real-world power grid with high IBG penetration. Case studies demonstrate that the proposed method outperforms state-of-the-art approaches by reducing computational burden while improving the proportion of stable states to over 95% in the operation simulations of the planned power systems.