With the increasing exploitation of wind energy, the dynamic performances of largescale wind farms (WFs) are vital for the system design and stability analysis. However, a detailed WF model considering every wind turbine system (WTS) makes the electromagnetic transient simulation time-consuming and computational-demanding. Therefore, a dynamic equivalent aggregation method is proposed for WTS with a full-scale power converter (WTS-FSPC) in this paper, in order to reduce the computational complexity meanwhile maintaining the same accuracy as the detailed model. To improve the accuracy of the equivalent collector system, the collecting strings are decoupled based on the power flow distribution. Then the WTSs are clustered according to the output AC voltage distributions using a Louvain K-means (LKM) algorithm, which innovatively determines the optimal solution automatically. Additionally, a centralized capacitor is set to represent the distributed capacitance of the collector system. Furthermore, to enhance the model adaptability during asymmetric short-circuit faults, the equivalent method for the zero-sequence collector system is introduced. Time-domain simulation results demonstrate the effectiveness and robustness of the proposed method in reflecting the WF dynamics. The case study in a 2-area system further verifies the superiority of the proposed method for the WF integrated power system stability analysis.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.