Intelligent setting of zinc dust dosage for a cobalt removal process based on estimation of zinc dust utilization ratio

Cobalt removal process involves the addition of zinc dust to remove impurity ions. Reasonable setting of zinc dust dosage is crucial in avoiding excessive zinc dust addition and guaranteeing the cobalt removal performance. However, due to the fluctuation of inlet conditions and the complexity of reaction dynamics inside the reactor, the precise determination of zinc dust dosage is challenging. In this paper, an intelligent setting method for zinc dust dosage based on estimation of zinc dust utilization ratio is proposed. Firstly, considering the accumulative effect of the large-scale metallurgical reactor, the reactant accumulation ratio (RAR) is defined. Then dynamic linear slow feature analysis (DSFA) is conducted using RAR and related process variables. The selected slow features describing the underlying process dynamics are used as the inputs of neural network to estimate the utilization ratio of zinc dust. Finally, the zinc dust dosage is determined according to the estimation result of utilization ratio. In order to verify the effectiveness of the proposed method, experiments were carried out using the actual production data of a cobait removal process. The experimental results illustrate the effectiveness of the proposed approach.