Abstract: Coal-gangue identification is a key technology for reducing the gangue content in raw coal during fully mechanized top-coal caving mining. Affected by the high dust concentration, low visibility, and limited operating space in underground working environments, existing identification methods based on a single feature such as images, multispectral and ray signals, sound, or vibration acceleration have difficulty achieving precise coal-gangue identification. To address this problem, vibration and sound signals were used as joint identification features, and a vibration-acoustic coupling model of coal-gangue impacting the tail beam of a hydraulic support was established using COMSOL Multiphysics software. The dynamic behavior and sound pressure frequency spectrum characteristics of coal-gangue impacting the tail beam under different gangue shapes and incidence angles were investigated, and the distribution patterns of vibration acceleration and sound pressure signals were obtained. The results showed that the stress, vibration acceleration amplitude, and sound pressure signal characteristics generated by gangue impacting the tail beam were all greater than those generated by coal. When spherical, cubic, and cylindrical coal-gangue particles impacted the tail beam, the maximum von Mises stress, vibration acceleration amplitude, and main frequency of the sound pressure decreased successively. As the incidence angle of coal-gangue particles increased, the contact force, peak vibration acceleration, and spectral centroid of the sound pressure signal during impact with the tail beam all decreased, and the response attenuation rate during gangue impact was higher than that during coal impact. These findings provide a theoretical basis for constructing a coal-gangue identification strategy based on the fusion of multi-feature vibration-sound pressure signals.