Abstract: In order to the vibration effect caused by foundation pit blasting excavation is comprehensively affected by many factors, the traditional empirical formula to predict the vibration speed is difficult to meet the current needs of blasting safety.Therefore, how to optimize blasting parameters and reduce blasting vibration effect is of great significance to ensure the safety of adjacent buildings.Based on the 400 groups of sample data obtained from the on-site blasting monitoring of a foundation pit project, this paper uses genetic algorithm to optimize BP neural network to predict the vibration velocity, and compares the vibration velocity prediction results of GA-BP neural network with those of BP neural network and Saab formula.The results show that the prediction accuracy of vibration velocity of BP neural network is significantly better than that of Saab formula, and the prediction accuracy of vibration velocity of BP neural network optimized by genetic algorithm is further improved.