Shield tunneling machine construction early warning method and device

The invention relates to a shield tunneling machine construction early warning method and device, and belongs to the technical field of tunnel boring, a full face rock tunnel boring machine (TBM) has the advantages of being high in construction speed, high in safety, good in economical efficiency and the like, and is widely applied to major projects such as expressways, railway transportation and urban subways. Due to the severe working environment, a real-time construction early warning method and device are needed. Aiming at the two fields of cutterhead detection and surrounding rock classification, the invention designs a method based on rock slag classification and surrounding rock integrity monitoring, and realizes tunnel construction monitoring and early warning. The two methods are both based on a convolutional neural network, so that the invention provides a convolutional neural network acceleration architecture based on OpenCL, and a deployment acceleration device for an early warning method is reali