摘要
采用正交试验研究萃取剂类型、萃取剂体积、分散剂类型、分散剂体积、溶液pH值、离子强度和萃取时间对水样中痕量十溴联苯醚(decaBDE)分散液液微萃取(DLLME)回收率的影响。结果表明,离子强度对萃取回收率(ER)的影响非常显著,而分散剂体积与萃取剂体积交互作用的影响不显著。通过极值法确定的decaBDE分散液液微萃取条件下的萃取回收率为88.24%。以正交试验数据为训练样本,以分散剂体积、萃取剂类型、pH值、离子强度及萃取时间为输入,萃取回收率为输出,建立了影响decaBDE分散液液微萃取的BP神经网络模型。模型检验样本预测输出值和试验值的决定系数为0.8734,表明模型可以预测水样中decaBDE分散液液微萃取的回收率。采用遗传算法工具箱对建立的BP神经网络模型进行优化求解,得到的优化分散液液微萃取条件下的decaBDE萃取回收率平均值为99.94%,比通过极值法确定的萃取回收率提高10%以上。
This paper takes as its aim to present the resuhs of our investigation on the effects of the extraction solvent, its volume, pH value of the solution, ionic strength, extraction time on the extraction of trace decabrominated diphenyl ether. For the research purpose, we have adopted the orthogonal design and the dispersive liquid-liquid microextraction method based on BP neural network model. The vari ance analysis of orthogonal experimental data has shown that the ionic strength would significantly impact the extraction recovery （ER）, but the mutual interaction of volumes of extraction solvent and the dispersive solvent proves to have little effect on ER. By extremum method, the extractive conditions can be obtained. Under such conditions, the mean extraction recovery of decabrominated diphenyl ether turns out to be only 88.24%. Then, we have established a prediction model for extraction recovery by using a dispersive liquid-liquid microextraction （DLLME） based on BP artificial neural network, in which the orthogonal experimental data were selected as the study sample with the due data. The determination coefficient between the output value predicted by the model and the experimental value has been worked out as 0. 873 4, which indicates that the model is reliable enough to forecast the extraction recovery in the dispersive liquid-liquid microextraction method. Based on the said model, genetic algorithm has been applied to optimize the extraction conditions of the dispersive liquid-liquid microextraction in the end. The validation experiments showed that the mean extraction recovery of the decabrominated diphenyl ether is 99.94% at the optimized extraction conditions. The results show that the trace decabrominated diphenyl ether can be well extracted at the optimized conditions by means of the DLLME from the environmental water sample with its extraction recovery being 10% higher than that of the extraction conditions obtained by the extremum method.
出处
《安全与环境学报》
CAS
CSCD
北大核心
2009年第4期62-66,共5页
Journal of Safety and Environment
基金
国家重点基础研究发展计划(2004CB418501)
关键词
环境科学
分散液液微萃取
十溴联苯醚
萃取回收率
BP神经网络
数学模拟与优化
environmental science
dispersive liquid-liquid microextraction
decabrominated diphenyl ether
extraction recovery
BP neural network
mathematic simulation and optimization
