Abstract: The safe and reliable operation of the hoisting system is very important for the production of the whole mine. Therefore, it is necessary to realize the single-point global visualization and virtual remote cooperative linkage control of the mining hoisting system, so as to solve the problem that the traditional multi-point video surveillance can only cover some key components and obtain incomplete information. To solving the problem, the digital twin framework of a mining hoisting system that with the function of monitoring, control and servicing is constructed by industrial sensing, artificial intelligence, rapid modeling, cloud storage and other technologies. Based on the framework, a multi-dimensional and multi-scale digital twin rapid modeling method is proposed. Firstly, a large scale geometric model of the mining hoisting system is built by 3D laser scanning technology, filtering, Poisson 3D reconstruction and other point cloud processing algorithms. Secondly, using industrial sensor network, PLC data reading and conversion technology, the behavior model of mining hoisting system under massive data is established. Lastly the fault knowledge model of the mining hoisting system is constructed using the database technology, domain expert knowledge and cases. The multi-dimensional and multi-scale digital twin rapid modeling experiment is carried out in a mine, and the results are as follows. The efficiency of the geometric modeling method has improved 93% compared with the traditional CAD modeling method and the modeling result has great similar to the real mining hoisting system. The behavior modeling method realizes the mapping of real entity behavior without adding new sensors and without shutting down, saves a lot of cost and has strong real-time performance. The driving Scripts are written based on Unity3D software to deeply integrate behavioral model, knowledge model and geometric model. Synchronization and deduction of behavioral model are carried out based on the component level model with high fidelity of the geometric model and it can realize the non-delay cooperative linkage between virtual and real systems. Meanwhile, relevant professional knowledge in the field is triggered by real-time behavioral data to assist decision making. The establishment process of the whole digital twin model takes into account the cost and effect, which will greatly improve the operation security and intelligence degree of the mining hoisting system.