Nonlinear Parameter Identification Method In Grinding Classification Process Of Bauxite Based On Differential Evolution

The prediction modeling for particle size of grinding circuit based on grinding mechanism is significant in realzing optimizing control, increasing grinding efficiency. In the continuous grinding-classification circuit, the wear of ball and the sand of spiral classifier have a great influence on particle size of grinding circuit. In this paper, the prediction model for the law of steel ball wear and the online prediction model for classification efficiency of spiral classifier are proposed on the basis of theories, and an improved differential evolution algorithm is used to identify parameters in those models.(1)Unconstrained nonlinear parameter identification method based on an improved differential evolution algorithm and constrained multi-objective model of nonlinear parameter identification method based on based on an improved differential evolution algorithm are established to study the nonlinear model parameter identification problems in grinding-classification process of bauxite.(2) An improved prediction model for the law of steel ball wear is studied on the basis of impact wear, abrasion wear and corrosion wear. An improved differential evolution is adopted to identify the parameters of the prediction model based on the ball wear data which are obtained by experiments in different conditions of grinding. The simulation results show that the variation of steel ball diameters can be simulated and predicted effectively based on the prediction method of the law of steel ball wear in ball mill, and the prediction model can be also used as a calculation basis in adding steel balls reasonably into the ball mill.(3)The existing mathematic models of spiral classifier are hard to realize online prediction of classification efficiency, combining the particle size distribution function, the mathematical model of spiral classifier and the RBF nerve network prediction model, the online prediction model based on constrained multi-objective nonlinear parameter identification method is proposed to study the efficiency of spiral classifier,and the constrained multi-objective model of nonlinear parameter identification method based on an improved differential evolution algorithm is used to identify the model parameters. The results based on online classification efficiency prediction model is slightly inferior to the results base on RBF nerve network offline prediction model, but the online prediction meets accuracy for the grinding-classification process of bauxite and can provide effective operation guidance for control and optimization of the production process-Figures (40), tables (21), references (66).