Hybrid Modeling And Model Correction Of Hydrometallurgy Copper Extraction Process

As one of the extractive metallurgy of the two technologies,compared with pyrometallurgy,hydrometallurgy is more suitable for smelting low-grade mineral resources.The solvent extraction technology simplifies the expensive solid-liquid separation step,leading to a promotion of a rapid development of hydrometallurgical processes.Currently,the extraction craft in hydrometallurgy in our country still remains in the condition of off-line analysis,experiential adjustments,and manual controlling,which has become a bottleneck of hydrometallurgical industry development.Based on deeply analyzing the characteristics of copper extraction production process,using hybrid modeling method,which is composed of mechanism and data modeling methods,this article carries out the modeling work of hydrometallurgical copper extraction process roundly and systematically.The main researches are summarized as follows:(1)The principles and process of copper extraction are introduced in detail,and the mass balance relations,a dynamic mechanism model of copper extraction process is modeled according to the relationship of material balance,and then,through simulation,the factors affecting the raffinate copper ion concentration is analyzed;(2)A serial hybrid model of copper extraction process is established,which is composed of a mechanism model with unknown parameter(equilibrium concentration)and a model which identifies the unknown parameter.The equilibrium concentration is identified by the data model using the least squares support vector machine(LSSVRM).Considering that the equilibrium concentration in practice is unpredictable,we estimate each value of equilibrium concentration corresponding to each set of inputs using Tikhonov regularization method firstly,this method can greatly reduce the influence on estimation results caused by concentration measurement noise;then,the serial hybrid model is built by the mechanism model and the data model based on LSSVRM;finally,through simulation,based on the prediction of raffinate copper ion concentration,the validity of the hybrid model is verified.(3)In the copper extraction process,due to the change of operating instructions and parameters,or the change of the raw material quality,and equipment wear,there may be a larger change in process characteristics.A model correction is established based on model performance assessment.Firstly,an error Gaussian mixture model(GMM)is developed to describe the probabilistic characterization of model prediction error,the hybrid model performance is assessed by statistics of the error distribution as well;then,the model correction strategy is carried out based on the model performance assessment;finally,through simulation,the validity of the model correction strategy is verified.