Abstract: The peak value of blasting vibration in open pit mine is the main index to evaluate blasting effect. In the scene of coal and rock interbedded blasting in open-pit mine, aiming at the problems that the existing prediction methods of blasting vibration peak value are difficult to achieve ideal prediction results, resulting in unreasonable design of blasting parameters and initiation network, a prediction model of blasting vibration peak value based on integrated particle swarm optimization support vector machine algorithm (GRA-EPSO-SVM) with grey correlation degree feature selection is proposed. Based on the coal and rock blasting in Yuanbaoshan open-pit coal mine under different occurrence conditions, hole spacing, row spacing, hole depth, maximum charge in single section, minimum resistance line, blast center spacing, elevation difference and peak particle vibration velocity were selected as input parameters, and grey correlation analysis (GRA) was used to filter redundant factors affecting peak blasting vibration velocity (hole depth, maximum charge of single section, minimum resistance line, peak particle velocity); using integrated particle swarm optimization algorithm (EPSO) to optimize the key parameters C and g of SVM algorithm, and inputting the parameters into GRA-EPSO-SVM model for evaluation. The results show that GRA-EPSO-SVM combination algorithm is more accurate than improved Sadovsky formula and SVM in predicting and actual values, and the average error is reduced by 15.3% and 106.8% respectively. The prediction accuracy is higher and the peak value of blasting vibration in coal and rock interbedded in open-pit mine can be predicted more effectively, which provides help for safety control of blasting construction in open-pit mine.