Wind farm integration in large-scale power systems for performing stability assessment requires significant modelling efforts and high computational time. In these cases, wind farm clustering is used to simplify the simulation efforts, but still, it requires a large number and composition of clusters to represent the abrupt change in wind speed and direction. A probabilistic clustering approach could be useful in such a case, which can identify the most probable cluster(s) in a wind regime within a timeframe, such as one year. This paper has presented a probabilistic clustering framework to represent the most recurring aggregated wind farm model throughout the whole year by implementing four clustering algorithms, namely (i) K-means, (ii) hierarchical, (c) fuzzy c-means, and (d) DBSCAN (density-based spatial clustering of applications with noise). The performance of the aggregated wind farm model has been compared with the detailed wind farm model in assessing the small-disturbance, frequency, and voltage stability of a power system. The simulation results show that the aggregated equivalent models (as identified by the probabilistic clustering approach) present the same level of accuracy while performing the simulation 10-times faster. This simulation efficiency could be very useful for performing dynamic studies for large-scale power systems.