Noise auto identification and de-noising method of coal-rock weak electromagnetic signals

When coal and rock break,they can produce electromagnetic signals,and the analysis of their characteristics is very important for the accurate prediction of coal rock dynamic disasters. However,when the signal is weak,the in- terference factors will greatly affect the result analysis. Thus,based on the analysis of noise sources and their respective characteristics of coal rock electromagnetic signals under laboratory conditions,the methods of loop band stop filter, mean filtering method based on white noise statistics and empirical mode decomposition (EMD),and periodic feature recognition along with band stop filter based on the frequency characteristics of modal component auto correlation func- tion,etc. ,were put forward to auto-identify and de-noise the different noises. In addition,a method of noise to noise ra- tio (NNR) was proposed to evaluate the de-noising effect in the case of unknown signal source. The result shows that the NNR of briquette electromagnetic signal by using the de-noising method developed in this study is only 0. 136 6, which is obviously better than the method of EMD or WAVELET. The result quantitatively indicates that the de-noising method has a good effect in the application of coal rock electromagnetic signal de-noising.