Dynamic equivalent modeling method for offshore wind farms based on fault characteristic multi-layer clustering

To effectively address the problem of low comprehensive simulation accuracy of dynamic equivalent models for offshore wind farms under diverse fault conditions, this paper proposes a dynamic equivalent modeling method for offshore wind farms based on cc (LVRT) fault electrical characteristics, control capability and AI algorithm three-layer clustering (FCAC). First, the wind turbines are divided into groups at the first layer based on the terminal voltage drop depth during faults, so as to incorporate the influence of turbines not entering LVRT on the external fault characteristics of the wind farm. Second, for the turbine group that have entered LVRT, a second-layer grouping is performed according to the difference in active power output during faults, aiming to integrate the impact of current inner-loop limiting on the external fault characteristics of the wind farm. Finally, the deep embedded clustering (DEC) algorithm is adopted for the third-layer grouping of the low-output LVRT turbine group, so as to fuse the influence of dynamic characteristic differences among turbines within the same LVRT control condition group on the external characteristics of the wind farm, and the final grouping results are obtained. The research results show that, as the proposed FCAC simultaneously accounts for the influences of grid voltage drop degree, LVRT control differences, and dynamic differences among turbines within the same LVRT control condition group on the external fault characteristics of offshore wind farms, it effectively improves the comprehensive adaptability of the dynamic equivalent model for offshore wind farms. Simulation results verify the correctness of the theoretical analysis.