A Signal is Worth 8 Words: Coal Mine Mechanical Fault Diagnosis with Transformers

Coal mine mechanical equipment runs in the environment of high dust and high humidity for a long time, and it is very easy to occur equipment failure, which is a great potential danger for the safe production of coal mines. Fault diagnosis is important for the safe operation of coal mine machinery. Currently, most of the research on fault diagnosis of mechanical equipment utilizes the attention mechanism in Transformer combined with CNN to extract important features, ignoring the ability of Transformer model to extract semantic relationships for long-term time series. Aiming at the above problems, this paper proposes a way to slice the signals of mechanical equipment into Patch sequences, and then use the Transformer model to learn the semantic relations between the sequences; finally, the fully connected layer is utilized for fault diagnosis. The effectiveness of the proposed method is verified by the Case Western Reserve University Bearing Data Center (CWRU) dataset and the equipment dataset collected from underground coal mine, and the experimental results show that the Transformer model that learning the semantic relationship between signal segments can realize the high-precision fault diagnosis results.