Abstract: Aiming at the problem of MEMS vector hydrophone denoising, a new combined EEMD-WT-SSA algorithm is proposed by combining the combination of set empirical mode decomposition（EEMD）, wavelet threshold denoising（WT） and singular spectrum analysis（SSA）. Firstly, the noise-containing signal is decomposed into a series of intrinsic mode functions（IMF）, then the high frequency and low frequency are distinguished by continuous mean square error criterion（CMSE）, the threshold value of high frequency signal is de-noising, and then the low-frequency signal is reconstructed. Finally, the reconstruction signal is recovered by singular value analysis method to obtain the target signal. The signal processing of simulation and lake experiment shows that the proposed algorithm has obvious advantages over EEMD and EEMD-WT in the aspect of SNR and MSE.