The distribution characteristics of blasting pile is an important index indicator to evaluate blasting effect.In view of the inadequacy of the current direct and indirect methods of measuring fragment size of the blast pile,a spatial distribution measurement method for adaptive stratification of the blast pile is proposed.It uses the GA-LSSVM model to predict the shape parameters α and β of the Weibull function and sets multiple prediction points to predict the three-dimensional blasting pile morphology.By converting and fusing the parameters of the Kuz-Ram fragment prediction model,a distance prediction model of blast pile stratification is established and applied to the Weibull-GA-LSSVM model to achieve an automatic stratification of the blast pile.Through field application,the stratification design is continuously optimized for the best stratification position to realize the adaptive stratification.The results show that:(1)the Weibull-GA-LSSVM model can accurately predict the morphology of the blast pile with a good stability that the average relative error of the prediction results of the maximum forward distance of the blast pile is only 5.6% and the relative error of the prediction results of the looseness coefficient is mostly around 9%.(2)The Kuz-Ram-based blast pile stratification model can reasonably output the layer distance and number before blast,which ensures the shoveling efficiency after blast.(3) The optimal layer distance formula is derived to achieve the adaptive stratification of the blast pile,and the measurement accuracy of the fragment size distribution of the blast pile is significantly improved,which is closer to the overall fragment size distribution.