一种低惯量电力系统中快速频率响应储备定容的在线方法

Rising frequency instability issues due to the phasing out of the synchronous generation capacity and the growing share of non-synchronous sources are creating concerns for power system security. The increasing volatility of system frequency due to diminishing system inertia and the inability of slow-acting reserves to contain the frequency decline have necessitated the procurement of the fast-frequency response reserve (FFR). Although such reserves can be procured from numerous sources that can deliver reserve power within seconds, quantifying such reserves is the immediate bottleneck. To address this issue, an online framework is proposed to size the FFR that can be obtained by existing solar photovoltaic plants. A machine learning-based regression model has been developed in the proposed framework to predict RoCoF and frequency nadir in varying system conditions and to assess system frequency security. Reserve distribution strategies that highlight the impact of network impedance and reserve delivery location on the overall improvement in frequency have been analyzed. Based on the system frequency security and the network information, an electrical distance-based clustering approach has been developed to avoid the excess procurement of the FFR. Case studies demonstrate that the proposed framework can effectively achieve the desired security with comparatively lower FFR.