摘要
针对Cu^(2+),Co^(2+),Zn^(2+) 3种金属离子混合溶液的紫外可见分光光度法(UV-Vis)光谱重叠而难以检测的问题,提出了一种基于改进型Monte Carle无信息变量消除(MC-UVE)方法的多金属离子浓度检测方法。在MC-UVE的基础上引入指数衰减函数(EDF),提出了一种改进的MC-UVE方法,并基于该方法对料液的紫外-可见光谱进行波长筛选;然后,利用筛选出的波长建立PLS模型并进行各组分浓度检测计算;最后,分别对基于MC-UVE方法和改进型MC-UVE方法 PLS模型的计算结果进行对比分析。结果表明:改进型MC-UVE方法可筛选出对模型贡献度高的变量,基于该变量选择方法的PLS模型精度高。
A polymetallic analytical method based on an improved Monte Carlo uninformative variable elimination(MC-UVE) was proposed to determine ion concentration by ultraviolet visible(UV-Vis) spectra of Cu(2+), Co(2+) and Zn(2+) mixture solution. The improved MC-UVE, developed by incorporating exponential attenuation function(EDF), was applied to selecting wavelength of UV-Vis spectra. And the partial least squares calibration model was built on concentrations of polymetallic components at the selected wavelength. Experimental results showed that the improved MC-UVE could select higher contributing model variables anddevelop more precise PLS model than MC-UVE.
出处
《化工学报》
EI
CAS
CSCD
北大核心
2017年第3期998-1004,共7页
CIESC Journal
基金
国家自然科学基金重点项目(61533021)
国家创新研究群体科学基金项目(61621062)~~