Abstract: In the process of fully mechanized caving, it is an important direction to improve the recovery rate of top coal and reduce the gangue rate of coal. In this paper, the principle and process of intelligent coal discharging process are systematically analyzed. It is pointed out that the key of intelligent top-coal caving process is coal and gangue recognition. This paper describes the principle of intelligent top-coal caving technology based on vibration signal detection, audio and video detection and ash online detection. This paper describes three key technologies for precise control of intelligent top-coal caving. By installed in the front bracket UWB through penetrating radar measurement top-coal thickness as the benchmark of coal quantity, the coal gangue identification based on vibration acceleration sensor is used to identify the coal gangue, with 9.5 KHz to 10.5 KHz range to be an effective signal for the termination of the coal gangue down benchmark, insert by tail beam in sublevel caving support and suet cylinder installation magnetostrictive stroke sensor precision control of coal body movements. According to the practice of fully mechanized caving in the 12309 working face of Wangjialing coal mine, the extraction rate of top coal reached 91% and the coal gangue content rate reached 8% by using the intelligent coal drawing process, which increased by 3% and decreased by 2% respectively compared with the manual coal drawing method of "see gangue close the window". Overall recovery and release efficiency increased by about 20%.The study can provide reference for intelligent top-caving in fully mechanized caving face of thick coal seam.