A Deep Learning-Based Wear Grading Method for PDC Drill Bit Composite Piece

The composite piece is the core cutting unit of the PDC drill bit. It directly affects the rock breaking efficiency and service life of the drill bit. Therefore, it is particularly important to grade the wear of the composite piece. This paper proposes a PDC drill bit composite piece based on deep learning. Wear rating method. First, the target detection network YOLOv7 is used to extract a single composite piece. Based on YOLOv7, this paper introduces depth-separable convolution, SimAM attention mechanism, SPPFCSPC, and K-means +  + clustering algorithm to improve the network accuracy and obtain Speed ​​improvement. Then, the composite piece area is semantically segmented through the U-Net network. Finally, ellipse fitting and grading are performed on the segmented composite pieces. Through experimental verification, the algorithm used in this article has greater advantages than other algorithms and can complete the task of composite piece wear grading.