Abstract: As a marine geological phenomenon caused by fluid overflow, submarine cold seep has gradually become a research hotspot in the context of increasingly advanced marine exploration technology. In the long-term monitoring of micro-seismic records in a cold spring area, there are a large number of micro-seismic signals of different types and characteristics. To intuitively and accurately describe the typical laws of each type of signal, it is necessary to extract common characteristics from a large number of micro-seismic signals. Therefore, the statistical classification and feature analysis of signals is an important part of the study of submarine cold seep activities. Starting from the micro-seismic signals and short disturbance signals related to the cold spring fluid identified in the study area of the “Haima” cold spring, the general characteristics of the two types of signals in the time domain and frequency domain were extracted by principal component analysis. For the short disturbance signal, frequency-attenuation (f-g) features were extracted from the first principal component record by spectrum analysis, on which the simulated single-frequency harmonic signal was compared with the actual record. Results show that this type of signal is composed of multiple sets of harmonic signals with different f-g components. Comprehensive use of the characteristics of the micro-seismic signals in the cold spring area could provide theoretical guidance for the follow-up study of the spatio-temporal response and generation mechanism of micro-earthquakes in a cold spring area, and a technical support for the study and evaluation of the development state of the cold spring system and the development of hydrate.