Abstract: In order to accurately identify the shape of the leaks for the in-orbit spacecraft, we propose a method based on the wavelet packet energy spectrums and the support vector machine. In this paper, firstly, the characteristics of the frequency spectrum of the holes of three typical shapes, namely, the circle, the rectangle and the equilateral triangle, are analyzed, respectively. Then the wavelet packet energy spectrum is used to extract the characteristic values of the leakage signals from different leaks. Finally, these characteristic values and the support vector machine are integrated to build a model for identifying the shape of the leaks. Experiments are carried out to validate the above method, leaks of different shapes are divided into the group A and the group B, where the group A is for training the identification model, and the group B is for testing the model. By this method, a total identification accuracy of 95.9% can be reached eventually with a proper number of decomposition layers.