Arsenic Salt Purification Addition To Copper Process Copper Ion Concentration Prediction Model Research And Application

As the first part of zinc electrolyte purification using arsenic-activated process, section of removing copper is to reduce the copper icons concentration to a required range,eliminating the negative impact to the next purification part and ensuring the normal follow-up process.In order to effectively control the copper concentration from the exports,the real-time detection values are essential.However,the copper concentration values are obtained by artificial test,which is very delayed and inaccurate, resulting unreasonable zinc addition and subsequent trouble, increasing production costs and energy consumption. Thus, it is of great theoretical and applicable significance to conduct research on。With deep analysis of technology and reaction mechanism of copper removal,a mechanism model is set up, which follows chemical reaction kinetics and material balance principle and predicts copper ion concentration of the export solution. Based on the mechanism of displacement reaction,this model points out the main impact factors of copper ion concentration.However, because of the complexity of the reaction mechanism,assumptions and simplification of modeling,it is hard to meet the requirement for prediction accuracy. In this case, based on support vector machine (LS-SVM), a predictive export copper ion concentration model is established by analyzing a large number of historical production data and and combining principal component analysis (PCA) principle. This model can describe the relationship among the training samples, while the predictive accuracy will decline and tracking speed will decrease under conditions of unstable conditions or large interference. Considering the global fitting ability of mechanistic models and the highly nonlinear mapping ability of least squares support vector machine model, an integrated modeling idea is introduced/which designs a control and coordination device, with production stability conditions as the criteria, intelligently integrating the two models, and establishing an export copper ion concentration intelligent integrated predictive modelFinally, the mechanism of prediction model, LS-SVM model and intelligent integrated predictive models are applied to the actual copper removal process.Based on comparision in terms of accuracy and robustness of prediction performance of each model, it shows that prediction of the intelligent integrated model is more accurate and more stable than that of the mechanism model or LS-SVM model (single model). As a result, optimal control of zinc addition is achieved which guides and ensures favorable conditions for the subsequent progress.