Based on the application of the classification control approach, a generalized algorithm for optimization of beneficiation processes is proposed. The results of computer modelling of the classification optimization process on the example of real indicators of magnetite quartzite beneficiation are presented. The results of classification and evolutionary optimization procedures are compared. It was concluded that the proposed intelligent classification method is able to determine the vector of settings and predict the TP beneficiation with satisfactory accuracy. It is confirmed that the developed algorithms and control principles can be applied to determine the required parameter values in modern ICS.