Application and research of FCOT based on DeepSeek in the mining field

Aiming at the problem of difficult knowledge precipitation and high learning cost of new employees in the fault maintenance of mining equipment, this paper proposes a Fault Chain of Thinking (FCOT) intelligent analysis method based on DeepSeek large model.Firstly, in order to solve the problem that traditional fault tree analysis relies on manual experience and equipment diagnosis and maintenance knowledge is difficult to precipitate, through the complex reasoning ability of the DeepSeek large model, a multi-level fault logic chain is constructed to realize the automatic extraction of knowledge, and the training set based on mining equipment faults is obtained. Secondly, the model distillation algorithm is used to realize the efficient transfer of expert experience, so that the small model also has deep thinking and expert ability. Then, aiming at the technical pain points of the traditional fault tree top-down analysis method with high complexity and difficult practical operation of new employees, a dynamic interactive module was built based on the distillation model and RAG technology to realize the rapid location of equipment faults in the form of questions and answers. Finally, through horizontal and vertical dual-dimensional comparative experiments, the effectiveness of our method in fault diagnosis scenarios was verified, providing a new technical path for realizing intelligent operation and maintenance of mining equipment.