A General Equivalent Modeling Method for DFIG Wind Farms Based on Data-Driven Modeling

Abstract

To enhance the stable performance of wind farm (WF) equivalent models in uncertain operating scenarios, a model-data-driven equivalent modeling method for doubly-fed induction generator (DFIG)-based WFs is proposed. Firstly, the aggregation-based WF equivalent models and the equivalent methods for aggregated parameters are analyzed and compared. Two mechanism models are selected from the perspective of practicality and complementarity of simulation accuracy. Secondly, the simulation parameters are set through two sampling methods to construct a training database. Next, the whole fault process is divided into five phases, and the weight coefficient optimization model is established according to the data-driven idea to achieve the adaptive configuration of the weight. Finally, the electromechanical transient simulations of the power systems with a DFIG-based WF is carried out by using the MATLAB/Simulink platform. Compared with the detailed WF model, the simulation time of the WF equivalent proposed in this paper can be significantly reduced by about 87%, and simulation results show that the proposed method can effectively improve the adaptability of the WF equivalent model in different wind scenarios and voltage dips.