Research On Prediction Method Of Cu Component Content In Hydrometallurgy Copper Extraction

Hydrometallurgy extraction process, as a metallurgical process, has the advantages such as dealing with low-taste of mineral raw materials, preventing environmental pollution, and so on. As a result, it is widely used in the metallurgical process. In hydrometallurgy copper extraction process, existing online detection equipments are difficult to get the Cu component content through online testing in the actual production process for the disadvantages such as huge investment, poor reliability of continuous operation, and low accuracy.Soft sensing technique is an effective way to solve this problem. Its basic idea is that we can choose some easier test process parameters (auxiliary variables) for inference and estimation by establishing some mathematical relationships to achieve the function that software takes place of hardware (sensors) for those key process parameters (leading variables) which are difficult to measure or impossible to measure. At present, soft sensing technique has been widely infiltrated into various engineering fields and become one of the most important research areas in process control and process detection fields.The research background of this thesis is hydrometallurgy copper extraction process. The principle and process of hydrometallurgy copper extraction process are deeply researched, and auxiliary variables, such as organic phase flow, leaching solution flow, lotion flow, strip liquor flow, strip liquor concentration and feed acidity, are selected to establish the prediction model of Cu component content. Based on the research of principles of cascade extraction process and chemical balance, the static model of copper extraction process based on the principle of material balance is deeply studied, and the need for the establishment of the hybrid model is identified. Soft-sensing theory and modeling theory are studied. Then the applicability and the superiority of the hybrid modeling method are analyzed, and the structure of the hybrid model is identified. Weighted coefficients of LS-SVM which are determined with the robust method can reduce the influence of noises. For this characteristic, the parallel hybrid model of Cu component content in hydrometallurgy copper extraction is established by the mechanism model and the weighted LS-SVM model. The prediction accuracy of the model may be reduced by the various factors in production process. For this problem, hybrid model is corrected by a correction method which is composed of the error correction mechanism and the model update mechanism. The test result indicates that the modeling method is effective.