Abstract: In view of the problems of large number of equipments,complex position relationship,lack of collision detection means between equipments and difficulty of pre-warning in fully mechanized mining face of coal mine.Combined with Virtual Reality (VR) technology,Multi-sensor Information Fusion (MSIF) technology and Bounding Box Collision Detection (BBCD) technology,this paper puts forward a method of combining virtual ray and bounding box to realize collision detection and early warning monitoring between fully-mechanized mining face equipment in the process of remote control of fully-mechanized mining face equipment,which is based on virtual environment development platform.Firstly,the composition and realization of collision detection and early warning monitoring system for fully-mechanized mining equipment based on virtual reality technology are introduced in detail.Secondly,the principle and method of virtual ray collision detection are described in detail by taking the relationship between the roller of shearer and the guard plate of hydraulic support as an example.Then,according to the coal mining technology,the movement attitude between the equipment is analyzed,and the mathematical model of roller-guard collision detection is established,and the collision detection and pre-warning monitoring between the shearer drum and the hydraulic support guard plate are realized efficiently.Finally,the experiment is verified by the virtual remote control platform of fully-mechanized mining equipment.By comparing the virtual ray detection results with the actual measurement results in the experiment,the maximum error of the collision detection results is within 49.6 mm,which meets the requirements of the actual working condition.The experimental results show that the virtual ray detection method can efficiently identify the collision between devices.It is of great significance to realize the abnormal detection of the fully-mechanized mining face by setting different thresholds to realize the pre-warning and monitoring of the inter-device classification collision.This method will be possible to solve the collision between fully mechanized mining equipment and other abnormal real-time monitoring.