基于知识与数据驱动融合Koopman方法的双馈感应发电机风电场频率支撑能力在线评估

Abstract The increasing integration of renewable energy in power systems causes a decrease in the frequency stability of the system. Consequently, renewable energy stations, such as wind farms (WFs), must possess adequate frequency support capabilities. To maximize the frequency support capability of the WF, it is crucial to determine the frequency support capability boundaries (FSCB) of the WF. Due to the uneven distribution of wind resources and complex operating states of wind turbines, accurate evaluation of the FSCB of the WF is challenging. To address this issue, this paper proposes a knowledge and data-driven fusion Koopman method to assess the FSCB of the doubly fed induction generator (DFIG)-based WF. The characteristics of FSCB are analyzed and a multi-dimensional indicator system is defined to precisely quantify FSCB at both theoretical and practical levels. To accurately calculate the defined indicators, a knowledge and data-driven fusion method based on Koopman-mixed integer linear programming (MILP) is proposed. The knowledge of WF frequency regulation structures is integrated to construct Koopman dictionary functions. This allows the training of historical frequency regulation data to obtain the global linearized Koopman operator for the assessment object. Subsequently, it facilitates online assessment results using real-time data. Case studies are undertaken on the four-machine two-area power system including a DFIG-based WF. The assessment error of the proposed Koopman-MILP method is within 2%, with an assessment speed nearly 10 times faster than conventional nonlinear methods. The proposed dictionary function, compared to the one without integrated knowledge, improves assessment accuracy by nearly 5 times. Additionally, it reveals the impact of frequency regulation strategies, safety operation constraints, and wind resources on FSCB. Simulation results validate the rationality of the proposed indicators, the accuracy of the assessment method, and the practicality of the assessment outcomes under various operating conditions.