The S transform(ST) time-frequency analysis technique is introduced to microseismic signal analysis field; the correlation between rock deformation-failure and microseismic signal frequency characteristics is investigated based on fracture scale. A large-scale seismicity observation is carried out in three hydropower engineering in Southwestern China; a significant amount of data is used as the statistical samples to minimize the influence of external factors on parameters calculating. Based on the statistical samples

the correlation between the source fracture scale and the frequency characteristics of microseismic events is analyzed. An engineering verification research is carried out combining with the monitoring data of microseismicity and multipoint extensometers in Houziyan hydropower station underground powerhouse. The research results show that the high frequency components decrease and microseismic signal displays low-frequency characteristic as the fracture scale increase; the microseismic high frequency components decrease at first and then increase during the deformation process of surrounding rock; and the frequency of microseismic events shifts from high band to a lower one before deformation. It can be believed as precursory information of deformation-failure that the frequency of microseismic events in concentrated area starts to decrease. Results can provide references for the stability analysis and hazards prediction of rock engineering based on microseismic monitoring.