A Voltage Sag Assessment Method for Distribution Networks Based on Key Control Parameter Identification of Distributed Generator

With the proliferation of distributed generators, the characteristics of distribution network faults transform, posing challenges to conventional voltage sag assessment methodologies. Specifically, traditional methods become less applicable due to accuracy issues, as low voltage ride-through (LVRT) requirements mandate that distributed generators no longer behave as constant voltage sources during faults. Furthermore, varying LVRT strategies among different units and the difficulty in acquiring their parameters hinder the establishment of precise equivalent source models. In response to these challenges, this paper proposed a novel approach for evaluating voltage sags in distribution networks based on identifying key control parameters of distributed power sources. The proposed method was initiated by constructing a parallel classification model that integrated explicit and implicit features, leveraging data-driven techniques to automatically identify crucial control parameters such as reactive power compensation coefficients during LVRT events. Subsequently, equivalent models of distributed generators under fault conditions were established based on the outcomes of parameter identification. On this foundation, Monte Carlo simulations were employed to carry out stochastic predictions of voltage sags. Simulation results affirmed the capability of the proposed method to discern control strategy parameters of various units accurately. Moreover, the parallel classification model demonstrated superior noise resistance and generalization performance compared to standard classification models. Consequently, the randomness in voltage sag prediction achieved through this framework exhibited a marked improvement over conventional methods, thereby validating its efficacy and potential in enhancing the accuracy of voltage sag assessments in evolving distribution networks.