关键词: 振动速度预测, 敏感性分析, 最小二乘支持向量机模型, 粒子群算法, 泛化能力

Abstract: Blasting seismic hazard is one of the most significant negative effects in the process of mining.Accurate prediction of peak particle velocity (PPV) is of great engineering practical significance to effectively prevent the instability and damage of buildings (structures) caused by blasting vibration.The blasting vibration test of openpit mine excavation was designed and carried out,the sensitivity analysis of the influencing factors of PPV was analyzed by the grey correlation analysis method,and the primary and secondary relationships between the influencing factors were determined.On this basis,the least squares support vector machine (LS-SVM) model is established to predict PPV,and the optimal parameter combination of regularization parameters and kernel function width coefficient in the LS-SVM model is determined by local optimization of particle swarm optimization (PSO).Finally,the prediction results of PSO-LSSVM model are compared with those of BP neural network model,LS-SVM model and traditional Sadowski formula.The results show that the fitted correlation coefficient (R2),root mean square error (RMSE),mean relative error (MRE) and Nash-Sutcliffe efficiency coefficient (NSE) of the PSO-LSSVM model for PPV prediction are 97.38%,2.68%,1.36% and 99.98%,respectively,and the PSO-LSSVM model has higher prediction accuracy and better generalization ability,which is feasible for mine blasting PPV prediction under multi-factor influence.

Key words: vibration velocity prediction,sensitivity analysis,least squares support vector machine model,particle swarm algorithm,generalization capability