An attention module integrated hybrid model for recognizing microseismic signals induced by high-pressure grouting in deep rock layers

• WM-ResNet50, integrating data enhancement, deep CNN, and CBAM, achieves 94.38% average recognition accuracy for grouting-induced microseismic signals, outperforming baseline models. • The hybrid model effectively classifies microseismic signals in deep rock layers, enabling accurate prediction of slurry diffusion ranges during high-pressure grouting. • Field and laboratory validations confirm that WM-ResNet50-classified MS sources match grouting records, demonstrating its practical utility in mining safety. • The study proposes a solution to address insufficient grouting diffusion under long-borehole conditions, validated via in-situ coring and XRD analysis.