Evaluation of Copper Mine Tailings Flocculation Behavior with PCA Biplot and GA-SVM

Abstract

— Flocculation treatment with polyacrylamide (PAM) was commonly applied for copper mine tailings treatment and recovery. In this paper, a complete evaluation method was established to effectively evaluate and predict the behavior of copper mine tailings flocculation with PAM. Firstly, the flocculation experiments were carried out with the self-designed dynamic test rig based on the orthogonal design. Furthermore, Principal Component Analysis (PCA) biplots were used for results evaluation. The range of optimal operating conditions were determined from PCA biplots based on 21 tests. Underflow concentration exhibited good correlation with flocculant dosage, overflow total suspended solids (TSS) displayed positive correlation with feed speed, and settling rate was positively correlated with slurry concentration. Biplot of the full data, including all the variables and targets, further indicated the relationships among the three target parameters. Finally, genetic algorithm-support vector machine (GA-SVM) was applied to predict the results of the parameters with various combination in the optimal range determined by PCA and the optimal operating conditions of copper mine tailings flocculation were determined ultimately. The proposed method system screened the flocculation operating conditions and predicted the flocculation behavior of copper mine tailings with PAM rapidly and accurately.