人工神经网络光度法同时测定四组份染料混合物  

Application of Artificial Neural Network on Simultaneous Spectrophotometer Determination of Four-components Dyestuff

作　　者：林生岭[1] 吴家建[1] 印厚飞[1] 朱炜峰[1] 曹旭[1] 侯莲莲[1] 

机构地区：[1]江苏科技大学材料科学与工程学院,江苏镇江212003

出　　处：《装备环境工程》2008年第4期19-22,共4页Equipment Environmental Engineering

基　　金：国家自然科学基金资助项目(20175008);江苏科技大学博士启动基金(2006CL0035)

摘　　要：介绍人工神经网络的原理,对典型神经网络系统做介绍,并对人工神经网络在光谱分析中的应用作解说。应用人工神经网络原理,以快速BP算法,对紫外可见吸收光谱严重重叠的四组分的染料溶液同时进行含量测定。在200-600nm的范围内,以12个特征波长处的吸收值作为网络特征参数,通过网络训练,酸性橙Ⅱ、酸性红B、甲基紫、酸性嫩黄的相对标准偏差分别为0.95%-2.30%,4种成分的回收率在96.3,%-104%之间。实验表明,该算法速度快,预测结果准确,可用于人工神经网络光度法定量测定光解废水中多组份混合染料。Principle and application of typical nerve network system in spectrum analysis were introduced. By means of artificial neural network and rapid back-propagation train algorithm. the four-components dyestuff was determined simultaneously, in which the ultraviolet-visible spectra overlapped badly. In the range of 200-600 nm. the absorbance at 12 wavelengths was taken as characteristic parameter of artificial neural network. The mean RSD of methyl violet, acid red B, acid orange Ⅱ , and acid light yellow is between 0.95%-2.30%, The recovery rate is between 96.3%-104%. The results showed that the algorithm is better in training speed and accuracy. It was concluded that the artificial nerve network based spectrophotometer is a good method for determination of multi-components dyestuff waste water.

关 键 词：人工神经网络 紫外可见分光光度法 染料 同时测定 

分 类 号：O657.32[理学—分析化学]