The invention discloses a 
voltage sag detection method for S-transformation on the basis of time-frequency gathering characteristic criteria. The method includes the first step of obtaining 
voltage sag disturbance signals to be detected generated in a power distribution network, the second step of conducting analog-
digital conversion on an obtained 
voltage sag disturbance 
signal model, the third step of conducting optimized selection on 
gaussian time-frequency 
window width adjustment parameters in an improved S-transformation method according to time-frequency gathering characteristic criteria, substituting the 
gaussian time-frequency 
window width adjustment parameters selected according to empirical values, and conducting characteristic extraction, detection and analysis on digital signals to be detected in the second step through the improved S-transformation. Selection of condition parameters of a 
gaussian window function is optimized through the constraint condition that the value of a time-frequency gathering characteristic expression is maximized when the signals are analyzed through the improved S-transformation, and the improved S-transformation is applied to characteristic extraction, detection and analysis on voltage sags. The 
voltage sag detection method for S-transformation on the basis of time-frequency gathering characteristic criteria has the higher detection accuracy degree and 
signal denoising capacity on transient-state 
electric energy quality signals.