论文部分内容阅读
用模拟数据和实验数据研究了人工神经网络（ＡＮＮ）方法用于解析色谱重叠峰的可能性，以二甲苯异构体及丙酮－异丙醇重叠峰为例对网络结构进行了优化，提出了两种简单易行的采集数据的新方法和模拟非正态峰的数学模型．结果表明，对不易分开的组分，只要训练集及测试集条件一致，大部分实验数据预测结果满意，用这种方法对色谱峰进行定量，结果可靠，避免花费大量时间寻找色谱分离最佳条件
The artificial neural network (ANN) method was used to analyze the possibility of overlapping chromatographic peak with the simulation data and the experimental data. The network structure was optimized by taking xylene isomers and acetone-isopropanol overlapping peaks as an example, Two Simple and Convenient Data Collection Methods and Mathematical Models to Simulate Nonnormal Peaks. The results show that for the components that are not easily separated, as long as the training set and the test set have the same conditions and most of the experimental data are satisfactory, the chromatographic peak can be quantified by this method and the result is reliable. It avoids spending a lot of time searching for the best conditions for chromatographic separation