To improve the prediction accuracy of blasting vibration in open-pit mines and mitigate the hazard risks to surrounding buildings(structures), a blasting vibration prediction model(DBO-BP) based on the Dung Beetle Optimizer(DBO) for optimizing the parameters of a BP neural network is proposed. Taking the blasting project of the Biesikuoduke Open-pit Mine in Hami, Xinjiang as an example, Grey Relational Analysis(GRA) is applied to evaluate the correlation between seven key parameters, including minimum burden, distance to the blast center, powder factor, stemming length, elevation difference, delay between rows, maximum charge per delay, and the radial, tangential, and vertical components of Peak Particle Velocity(PPV) of blasting vibration. The DBO algorithm is introduced to optimize the weights and thresholds of the BP neural network, constructing the GRA-DBO-BP model to predict the radial, tangential, and vertical PPV of blasting vibration. A comparative analysis of the prediction results from the GRA-BP and GRA-DBO-BP neural network models verifies the superiority of the GRA-DBO-BP model. To further verify the universality and reliability of the proposed method, the slope rock blasting project at the Nan Open-pit Coal Mine of Xinjiang Tianchi Energy was selected as the research object for validation. The results demonstrate that GRA can effectively assess the correlation between blasting parameters and PPV, with the grey relational grade for all seven parameters exceeding 0.6. DBO optimization can effectively circumvent the tendency of BP neural networks to converge to local optima, thereby enhancing prediction reliability and accuracy. This study provides theoretical support and technical reference for vibration control and safe construction in open-pit blasting projects.