Abstract: At present, the determination of rock mechanical parameters is limited to the probe and data acquisition equipment, which can only be used in the laboratory environment. At the same time, the traditional ultrasonic transceivers have high requirements for the contact surface roughness. Therefore, there is a need to polish the rock surface, and to apply sound guide paste before the analysis. The aforementioned issues have greatly limited the scope of the application of ultrasonic measurement systems. In order to achieve long-term monitoring in rock boreholes, the sound wave emission and reception mechanism and the frequency domain characteristics of the ultrasonic signals in the rock were analyzed in this study. The wavefront first arrival was picked up by the AIC algorithm, and the wavelet denoising and the improved AIC method (NAIC) were utilized to find the exact wave arrival time, and the internal parameter measurement system based on dry coupling rock was studied. The ultrasonic signal transmission and reception, signal transient triggering, and high speed acquisition were achieved inside the rock, and the final result was wirelessly transmitted to the data recorder. Ultrasonic transmission parameters were measured for Φ50 mm×100 mm granite and sandstone rock samples. The measurement distances of Φ114 mm×280 mm granite rock samples were 150 mm and 200 mm with the use of NAIC, STA/LTA, MER and ETA algorithms. The data was processed and compared to the results of third party instrument measurements. The results show that the system based on dry coupling acoustic wave velocity measurement has good adaptability to different working contact surfaces, and the data error is within the allowable range of the specification. Wavelet denoising is suitable for ultrasonic receiving signals with less background interference and concentrated frequency range. In terms of picking-up, the other three algorithms except STA/LTA can effectively improve the pick-up accuracy after denoising. NAIC is the method with the most significant error reduction, and the average error is reduced from 1.21 ?s to 0.19 ?s.

Key words: dry coupling, digital acquisition, wavelet denoising, improved AIC algorithm, arrival time picking