In order to improve the reliability and accuracy of acoustic emission prediction technology in coal roadway driving, and to reduce the influence of working noise on monitoring signal and monitoring effect in coal roadway driving, roadway support and borehole construction, this paper adopted the method which combined stress loading experiment in laboratory and the field monitoring and analysis, and studied the noise processing method of acoustic emission monitoring signal in the heading face of coal roadway. The results shown that the frequency distribution of operational signals from comprehensive excavator driving, drilling anchor and drilling rig drilling ranged from 0 to 8000 Hz, and the main frequency was in the range of 500 Hz, 3000 ~ 5000 Hz, 7500 ~ 8000 Hz, had“W” shape distribution features. The width of main frequency band of AE signal was 0 ~ 500Hz during the failure process of coal and rock samples, the dominant frequency did not exceed 250 Hz. In combination with the characteristic difference between the acoustic emission signal in the rupture process of the laboratory coal and rock specimen and the acoustic emission mixed signal in the coal roadway excavation and the spectrum characteristics, we adopted the reasonable configuration of zero drift value and signal threshold value, and combined the band-pass filtering and RII filtering to establish the noise filtering method of the monitoring signal, which realized the extraction of the effective acoustic emission monitoring signal during the coal roadway excavation operation. The noise filtering effect was conformed to the reality. This paper provided a method support for real-time acquisition and noise filtering of acoustic emission monitoring signals, and brought high accuracy monitoring and early warning. It had certain guidance and reference significance.