Abstract: The major challenges facing the real-time structural health monitoring (SHM) application in the spacecraft domain involve the transmission and the storing of a huge amount of data. In this paper, we propose an approach of the semi-tensor product compressive sensing (STP-CS) to address this problem is the SHM. The STP-CS can be regarded as a generalized version of the traditional CS in its capability of reducing the dimension of a random observation matrix. Due to the sparsity of the deflection data, a random sampling scheme is adopted based on the STP-CS to minimize the amount of the field data. The relationships between different observation matrix dimensions and the reconstruction performance are determined. The experimental results demonstrate that even if the storage space of the observation matrix is reduced to 1/64 of the traditional CS, the reconstruction can still be realized with a sufficiently high precision. In addition, the reconstruction time also decreases with the decrease of the dimension of the observation matrix. Therefore, this method provides a new tool to deal with the challenge of the data transmission and storage in the SHM of spacecraft.