Prediction control research of cobalt lion concentration in Zinc hydrometallurgy based on NN and GM technique

Abstract

Considering the strong non-linearity and large time delay of purification in zinc hydrometallurgy in purification process, a prediction model of cobalt concentration combining neural network (NN) and grey model (GM) are proposed. In the key part II of the purification, because of the harmful impurities cobalt ion concentration can not be on-line measured, and the testing results is two hours later before the production situation, it is mainly through the production of excessive addition of antimony salts and zinc to remove cobalt ions. Under the premise of deeply analysis of the process II of product and purification technique and the relevant factors, a control method was brought forward in this article, which adopts the same dimension grey prediction method to forecast cobalt ion concentration, and then uses a neural network technique to compensate the error of grey forecast. In the end of the article, two simulated experiments of the neural network grey model (NN-GM) and the least square support vector machine (LS-SVM) were compared. The results of the simulation and production practice has proved that the NN-GM model can be so better to predict the cobalt ion concentration values, that played guiding role for the operation process in optimize the antimony salt and zinc addition.