In view of the difficulty in evaluating the oscillation stability of direct-drive wind farm integrated power system during fault recovery, this paper proposed an oscillation stability analysis method based on dynamic energy. And the proposed method can depict the path of dynamic energy within the system, quantitatively analyze the influence of each energy interaction behaviors on the system's stability, and identify the key factors inducing the system oscillation. Firstly, based on the switching characteristic of D-PMSG control strategy, the system is partitioned into low-order subsystems, and each subsystem's energy model is constructed. On the basis, the dynamic energy items that characterize the energy interaction between different subsystems are extracted, which can be used to quantitatively evaluate the influence of the interaction energy on the system's stability. Furthermore, a method for quantitative analysis stability is put forward, and the key factors inducing system oscillation instability are identified. Finally, the accuracy and effectiveness of the proposed oscillation stability analysis method are verified by the hardware-in-loop tests. From the tests results, during fault recovery, the system's oscillation stability can be accurately assessed by the proposed method, and the key factors inducing oscillations can be identified.