부산대학교 도서관

This article presents a data-driven methodology for impedance identification and online stability assessment of wind farm integration with modular multilevel converter-based high-volage dc (MMC-HVDC) transmission system, which is applicable for black/grey-box models in practice. In order to improve the impedance identification accuracy, a neural network (NN) optimization method based on the evaluation coefficient is proposed to train the wideband impedance identification models of wind turbine generator (WTG) and MMC. On this basis, the physics-informed data-driven method for online wideband impedance identification of a wind farm is then proposed by combining the NN-based impedance identification models of WTGs and physical topology information of wind farm, which has the advantages of low data requirements and better interpretability. Based on the identified impedances, the oscillatory stability of the wind farm-MMC interconnected system can be evaluated online using generalized Nyquist criterion. Finally, case studies are carried out to validate the accuracy of the proposed impedance identification and online stability assessment methods.