Abstract: In order to improve the precision and safety of equipment control in fully mechanized mining face, an intelligentization collaborative control system based on digital twin technology is proposed. The system integrates digital twin, 5G communication, AI intelligent analysis and other technologies to build a digital twin with fully mechanized mining equipment, geological environment, and personnel distribution as the core, achieving real-time monitoring, remote control, and autonomous cutting optimization of equipment. The system adopts AI inspection robots, combined with deep learning models such as YOLOv5, ResNet and others to achieve intelligent identification and abnormal early warning of downhole environmental parameters, equipment status, and personnel behavior. Three months of on-site testing at the 150202 fully mechanized mining face of Youzhong Coal Industry shows that under downhole high dust and weak light conditions, the identification accuracy rate of AI model remains stable at over 91%. After the system implementation, the work efficiency of stoping is increased by 39%, the number of downhole operators is reduced by 26%, and the advance amount of disaster early warning is extended to 60 minutes.