The invention discloses an
algorithm for separating micro-
seismic noise mixed signals by using an SVD-EMD
algorithm. The
algorithm comprises the following steps: (S1) preprocessing the micro-
seismic noise mixed signals by using a
singular value decomposition method SVD; (2) obtaining a
signal f (x, t) subjected to SVD preprocessing, carrying out EMD
decomposition on the
signal, calculating currentstep signals f0 (x, t) of all extreme points by initializing a residual function res (x, t), and then removing an average envelope value from the original
signal to enable the remaining part of signals f1 (x, t) to become signals of the next step; (S3) placing the current step signal f1 (x, t) and the previous step signal f0 (x, t) in a stop state, and repeating the screening process through an intrinsic mode function and a final remaining function; and (S4) utilizing Hilbert transform to obtain frequency characteristics of each intrinsic mode function decomposed through EMD, and calculatinga
correlation coefficient Ri so as to complete
signal reconstruction. By means of the scheme, the purposes of improving
microseism seismic phase recognition and realizing the optimal
P wave picking effect are achieved, and high practical value and promotional value are achieved.