The most critical problem in numerical simulation of the blasting vibration is how to calculate the blasting load accurately and how to apply it.The accurate application of blasting load is the primary problem. In this paper, the loading method of the wall of blasthole is optimized. In order to obtain accurate time of rise and fall of the triangle blasting load and guarantee the reliability of the numerical simulation,this paper proposed neural network analysis to obtain the reasonable time of rise and fall. Firstly, the optimal combination of the time of rise and fall is obtained by orthogonal experiment ,the forward modeling method was used to acquire the sample for train and for test.Then the BP neural network is optimized by particle swarm optimization algorithm to find out the nonlinear mapping relationship between the time of rise and fall and the vibration velocity of blasting observation point. Finally,this paper put the blasting monitoring information as input samples to conduct inverse analysis for triangle blasting load time history curve of the time of rise and fall.Compared with the literature[4-5],the absolute error of the result is reduced by 0.7819(cm/s),the relative error is reduced by 11.5324%.So it is concluded that the method is feasible.