Zinc Hydrometallurgy Reverse Antimony Salt Purification Sec Metal Ion Concentration Prediction Model

The reverse antimony trioxide purification process is the key process of zinc hydrometallurgy, the aim of which is to remove all sorts of metallic impurities by adding the zinc powder and antimony trioxide. Due to the complexity of reaction mechanism, the characteristic of long flow, the existence of measurement-delay, it is difficult to control and optimize the polymerization process. It is significant for achieving process optimal control and reducing the consumption of raw materials to make some research on the approaches of prediction of the multi-metal ion concentration.Based on the analysis of the techniques and influencing factors of the purification process, a predication model for the concentration of the cobalt and cadmium ions is constructed by combining with the electrochemistry theory, support vector machine (SVM) and intelligent integrated model theory. The major innovation research work and achievements included:(1) A mechanistic model based on reaction kinetics was established by analyzing the metallic ions replacement reaction mechanism. And the diffusion coefficient constant was identified by partial least squares (PLS). The model has been validated by industrial data, and the results show that the model reflects the industrial processes well. Nevertheless, the complexity of mechanism and the suppositions and simplifications during modeling make it hard to meet the need of practical industrial requirement.(2) Prediction method of the ion concentration was studied on the basis of least squares support vector machine (LS-SVM). Aiming at the parameter optimization problem in least squares support vector machine, a hybrid QPSO algorithm (HQPSO) for LS-SVM parameter selection was proposed to improve the learning performance and generalization of the LS-SVM model. Then the production data from a purification process of zinc hydrometallurgy was used to verify the model performance, the simulation results show that the model is with higher precision. However, LS-SVM model exsits low generalization ability with unstable conditions and interference.(3) Aiming at the problems existing in the mechanism model and LS-SVM model, an integrated model based on mechanism model and LS-SVM model was proposed to meet the requirements of practical industrial process. The intelligence coordinator adjusted the outputs of two models, so the organic integration was realized. The model was tested by the data from purification process in zinc hydrometallurgy, and the simulation results show that the model has better performance and higher precision, which offered refernece for optimizing operation of practical production process.