Abstract: In order to improve the blasting effect and decrease the payment with non-pillar sublevel caving mining method for lead-zinc deposit in Xinjiang.According to the special geological of mining area and the existing mining technology,an artificial neural network model for forecasting and controlling the mining cost was established to search for the best blasting parameters in the process of blasting.The experiment with the model predictions turned out that when explosives consumption was 0.85kg/m³,the row spacing was 1.75m,the bottom of the hole distance was 2.1m,caving step distance was 5.25m certain had the best performance.All in all,the BP neural network model can accurately optimize the blasting parameters,and provide a new idea for optimal design of mining blasting parameters.