Abstract: The locating algorithm of multi-channel microseismic monitoring system is generally based on the P-wave arrival time and equal velocity model. The P-wave arrival time directly influences the precision of positioning. However, due to the attenuation of earthquake wave propagation and background noise, some channels are involved in positioning with low SNR, and random picking of P-wave arrival time may lead to great positioning error and affect the actual application effect in mines. The source location method and application of the noise signal filtering procedure based on wavelet theory are introduced. According to the wavelet decomposition, given threshold and reconstruction of signal, the SNR of the original signals is enhanced, and the picking accuracy of the P-wave arrival time is significantly improved. Taking some roof collapse examples in a mine with outsize goaf, re-optimization of picking the P-wave arrivals time in low SNR channels improves the precision of positioning results from 47 ~ 94 m to 14 ~ 23 m, and the actual application effect in mines is greatly improved.