Abstract: In order to realize the multi-machine equivalence of doubly-fed induction generator（DFIG）-based wind farms, it is necessary to accurately distinguish the units with different dynamic characteristics. However, the existing research has not established the quantitative clustering index based on the different power characteristics of low voltage ride-through for units.Therefore, a multi-machine equivalent method for DFIG-based wind farm based on power characteristics of low voltage ridethrough is proposed. First, based on the typical low voltage ride-through control strategy, the mathematical description of output power characteristics for units is established. Then, a two-step clustering method is proposed by analyzing the difference of power dynamic characteristics among units: in the first step, all units are divided into two groups based on whether the steady-state active power during the fault can be restored to the value before the fault; in the second step, the clustering algorithm is used to further divide the cluster to reduce the equivalent error. In order to improve the adaptability of the equivalent method to different wind speed scenarios, the affinity propagation（AP） algorithm is introduced to implement adaptive clustering. Finally, based on 10 sets of measured wind speed scenarios, the power curves and their error levels of the proposed method and other methods are compared. Simulation and calculation results verify the effectiveness of the proposed method.