Abstract
Purified liquid is obtained in the last stage of the zinc-smelting purification process, and real-time detection of its trace metal ion concentration is beneficial in providing effective information for the purification of zinc hydrometallurgy. We added Cu(Ⅱ), Co(Ⅱ) and Fe(Ⅲ) standard solutions to the purified liquid obtained from zinc smelting based on the analysis system of Nitroso R salt to simulate the fluctuation of the actual ion concentration, and adopted UV-vis spectrometry to determine the concentration of metal ions. Then, we adopted the strategy of "separation-extraction-calibration" strategy to quantitatively analyze the concentration of trace metal ions, namely, Cu, Co, and Fe, simultaneously. Elite Variable Cluster and Improved Local Weighted Partial Least Squares (EVC-ILWPLS) algorithm are proposed for feature extraction and multivariate calibration model. In the approach, the most effective information variable clustering is filtered using forward elimination and backward clustering, and then ILWPLS based on variable weighted and sample weighted is used to determine trace metal ions concentration. Experimental results indicate that EVC-ILWPLS can not only select effective variables quickly and accurately, but also establish a reliable regression model for spectral data with complex multivariate relationships and strong collinearity. A comparison between the proposed and other conventional methods shows that this proposed strategy can determine the concentration of Cu, Co, and Fe with the lowest RMSE and MRE . Therefore, the developed method can be implemented as a promising method for quantitatively analyzing trace metal ions concentration, and it provides a guiding significance in the analysis and processing of multivariate collinearity spectrum data.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call