Abstract: In order to ensure that the flotation machine has sufficient aeration and can produce the static separation environment required for mineral flotation，a double impeller control system flotation machine is designed by combining the centrifugal impeller with the stirring impeller. Based on the previous research，by fixing the structural parameters of the centrifugal impeller of the double-impeller flotation machine，selecting the diameter and speed of the impeller of the double-impeller flotation machine as the in⁃ put factors，and the beneficiation efficiency of the phosphate rock as the output factor，the prediction model of the beneficiation efficiency of the double-impeller flotation machine was established，and the accura⁃ cy of the model was tested by samples. The results show that the established BP neural network model can accurately predict the beneficiation efficiency of double impeller flotation machine，and the relative error between the predicted value and the experimental value is generally less than 5%. The established prediction model of beneficiation efficiency can be used for optimal control and decision-making of flotation pa⁃ rameters of double impeller flotation machine，which can reduce the amount of test and save manpower， material resources and time.

Key words: BP neural network, double impeller flotation machine, beneficiation efficiency, predictive model