摘要
用人工神经网络解析复方磺胺甲噁唑片的紫外吸收光谱数据,达到同时测定各组分含量的目的.按L25(56)正交表设计,制备了25组标准溶液的混合液,将其吸光度数据和浓度数据作为人工神经网络的训练集.混合液中各组分的5个浓度水平分别为80%、90%、100%、110%、120%.预报集采用自制的模拟样品和市售的复方磺胺甲噁唑片的吸光度数据.网络的输入为各溶液在246～290nm间的吸光度,网络的输出为各组分的浓度.利用Bayesian规则化调整的BP人工神经网络处理数据.结果表明,人工神经网络紫外分光光度法预测模拟样品中的磺胺甲噁唑(SMZ)、甲氧苄啶(TMP)的含量,平均回收率分别为100.53%和100.91%,相对标准偏差分别为1.17%和2.79%.对市售的复方磺胺甲噁唑片中的SMZ、TMP的含量也能取得较好的预测结果.结论:人工神经网络紫外分光光度法可以快速、准确地测定复方磺胺甲噁唑片中组分含量.
Aim： To delve neural network spectrophotoinetry applying to content determination of compound sulfamethoxazole tablets. Method： The train set was consisted of 25 group standard mixture which prepared according to orthogonal layout. Five different concentration level in mixture was 80% , 90% , 100% , 110% and 120% , respectively. The predict set was absorbance data of simulated sample and compound sulfamethoxazole tablets. The input of network was mixture absorbance data and the output was component concentration. The absorbance data of compound sulfamethoxazole tablets was processed with BP network of Bayesian regularization. Result： Experimental result indicates that appropriate network parameter can help to determine simultaneously sulfamethoxazole （SMZ） and trimethoprim（TMP） in the simulated sample and compound sulfamethoxazole tablets. The average recoveries and the relative standard deviation of SMZ, TMP were 100. 53% , 1. 17% ; 100. 91% , 2.79% , respectively. Conclusions： Artificial neural networks spectrophotometry is simple and accuracy in the simultaneous determination of compound sulfamethoxazole tablets.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2005年第8期643-646,共4页
Computers and Applied Chemistry
基金
天津市自然科学基金资助项目(T39601)
作者简介
李宁（1975-），男，在读博士研究生，主要研究方向为现代药物制剂及药物分析．
