To address the quality issue of CFRP delamination defects during the drilling process of laminated assemblies involving carbon fiber reinforced polymer （CFRP） specimens and titanium alloy specimens， a CFRP delamination damage identification method based on convolutional neural networks was proposed. Drilling tests were conducted using process parameters selected through an optimization experiment designed to minimize the impact on delamination. Axial force signals during the drilling process were collected， and continuous wavelet transform was applied to obtain wavelet scale maps as input sets， with delamination factors serving as label sets to construct a convolutional neural network model. The results indicate a positive correlation between axial force signals and CFRP delamination damage levels， demonstrating the feasibility of monitoring drilling force signals via the convolutional neural network model to identify CFRP delamination damage. Three network architectures VGGNet-16， LeNet， and AlexNet are implemented， and comparative analysis reveals that the VGGNet-16 model achieve the best recognition performance， with an accuracy rate reaching 91.84%.