With the development and initial use of reusable spacecraft， online monitoring the structural responses， understanding the structural health status in real time， and effectively identifying the structural damage state have become important ways for reusable spacecraft to ensure its economy efficiency. In this paper， a multi-damage self-identification method is proposed for the online assessment of composite structural damages in reusable spacecraft. The method first achieves the high-precision imaging of multiple damages in composite structures based on guided waves. Then， a probabilistic mixture model is used to cluster the damaged pixels， where the pixel values with high damage probability in the monitoring area are matched with the Gaussian model components. This allows the damage status of the structure to be automatically evaluated based on the number， mean value， and covariance of the Gaussian model components. Finally， experiments are carried out to verify the proposed method. The results show that the method can effectively identify the number， locations， and directions of the damages.