Product Quality Control For Hydrometallurgy Cementation Process Based On Stacked LSSVM

China contains a large number of low-grade nonferrous metal resources. With the increase of national economy and the development of industrialization, making full use of mineral resources has become more and more important to the strategy of sustainable development in China. As one of the two extraction technologies of smelting, hydrometallurgy has the advantages on high percentage of recovery of valued metals, protection of the environment and achievement of the continuity and automation in the production process. Considering of these respects, hydrometallurgy is more suitable for recovery of low-grade mineral resource. Cementation process is an important procedure of hydrometallurgy, whose control is stay in the level of off-line analysis, experience adjustments and manual control now in China. These result in low productivity, high consumption and instability of product quality in the vast majority of hydrometallurgy enterprises, which restrict the development of China’s hydrometallurgy industry.This thesis aims at the control of the product quality in cementation process of hydrometallurgy. Based on deeply analyzing the characteristics of cementation production process, it is researched comprehensively and systematically on mechanism modeling, data modeling and optimization in a batch. The main researches are summarized as follows:1. Based on deeply analysis on the cementation process, and according to the principle of reaction kinetics, material balance as well as the filtration equation of filter press, this thesis establishes the mechanism model of zinc dust replacement process, and the steady-state characteristics are analyzed through simulation.2. To the problem of difficulty in applying the mechanism model to the industrial field directly, the data modeling method can be used to predict the quality of products. Combined with the characteristics of cementation process, the method of off-line modeling for batch process, which is based on stacked least squares support vector machine (Stacked LSSVM), is adopted. By selecting kernel function, determining instrumental variables and selecting Members data of the LSSVM model, data model of cementation process is established. And then the validity of the modeling method is verified by simulation.3. Due to the uncertainty and disturbance in batch of cementation process, the amount of zinc dust added will not precise enough and the end product quality may deviate from the expected quality. To solve this problem, batch optimization control is proposed. Based on the data reconstruction approach of principal component analysis (PCA), the gold grade is predicted online. The thesis proposes optimal control strategy which based on the genetic algorithm (GA). Through determining the optimization objective function and analyzing feasibility of the operating variables, solve the optimization problem to the amount of zinc dust added. Simulation indicates that this method can effectively calculate the amount of zinc dust added in the follow-up period on-line, which can also guide the production.4. Last, this thesis designs the platform that is used for prediction and optimization in hydrometallurgy cementation process, as well as introduces the structure and function of each part.