Abstract: Coal seam thickness is a key controlling factor for coal mine production and safety, as well as a core parameter for the construction of smart mines. Traditional detection methods have achieved some success in medium- and thin-seam coal mines; however, in thick and extra-thick coal seams, due to significant variations in thickness and the frequent presence of interbedded rock, differences in the parameters of the coal seam and the interbedded rock often lead to detection errors. Aiming at the problem of detection errors caused by intercalated gangue in extra-thick coal seams, this study proposes a dynamic elastic wave detection method based on coal-rock interface recognition. Through coal-rock wave velocity testing, non-zero offset seismic principles, and time-domain finite difference simulation, the reflection characteristics of seismic waves under different coal thicknesses, wave impedance differences, and geological backgrounds are systematically analyzed. The results show that: due to the large wave impedance difference at the coal-rock interface, the reflection signal is significantly stronger than that at the rock-rock interface; the coal seam thickness mainly affects the signal time delay, while the lithological difference has little impact, a 5-meter difference in thickness results in a delay difference of approximately 1 millisecond; the average error of the on-site industrial test at 7 detection points does not exceed 4.89%. The on-site industrial tests have also shown that when loose coal is present on the floor, striking the iron tray with a hammer yields detection results that more closely reflect the actual coal thickness. The study verifies the effectiveness and engineering applicability of this method in detecting the thickness of the top and bottom coal in extra-thick coal seams.