Research On Modeling And Correcting Methods In Gold Cyanide Leaching Process

As the most important production process in Hydrometallurgy,the quality index of Leaching process has a direct effect on the recovery of valuable metals.Establishing softsensing model between operation variables and quality indexes is the basis of the process quality index control.However,in the actual production process,due to the change of raw material properties and the sedimentation of large particles in pulp,the leaching process will undergo change in working conditions or slowly time-varying,which will lead to decline in model accuracy.Therefore,it is of great theoretical and practical significance to study how to effectively ensure that the model performance satisfies the production requirements under the condition that the characteristics of the leaching process change.First,this paper introduce the hydrometallurgical leaching process and the mechanism model of the gold cyanide leaching process.The simulation verifies the feasibility of the mechanism model which is taken as the actual gold cyanide leaching process.An overview of partial least squares(PLS)modeling methods is also produced,which is employed to establish a soft-sensing model for cyanide slag gold concentration.Second,in order to monitor leaching process working conditions or slowly time-varying in real time,this paper build monitoring models using model-based principal component analysis(MBPCA),which can monitor performance index of soft-sensing model.When the monitoring model detects changes in the characteristics of the leaching process,this paper proposes to a method for identifying changes in leaching process characteristics based on wavelet analysis and SVM.Which extract the features of the monitoring model statistics based on wavelet analysis theory and use these features as input to train the SVM classifier,so as to identify whether the gold cyanidation leaching process occurs working conditions or slowly time-varying.When it is recognized that the leaching process takes place slowly time-varying,the on-site staff is instructed to clean the leaching tank so as to eliminate the influence of slowly time-varying on the leaching process;when identifying in the working conditions of the leaching process,this paper introduces casebased and just in time(JIT)model correction method.The overall idea is to first match the similar model from the model casebase.If there is a similar model,the similar model is directly switched;if there is no similar model,the soft-sensing model is rebuilt based on the JIT method to realize the model correction,and added into the model casebase to update the casebase.Finally,simulation studies verify the effectiveness of the proposed method.