BP网络在类黄酮类天然抗氧化剂构性关系研究中的应用  

Application of Back Propagation Neural Network in the Quantitative Structure-property Relationship of the Flavonoid Natural Antioxidant

作　　者：何琴[1] 黄保军[2] 高建华[3] 

机构地区：[1]许昌学院化学系 [2]许昌学院表面微纳米材料研究所,许昌461000 [3].郑州大学化学系,郑州450052

出　　处：《天津师范大学学报（自然科学版）》2007年第1期8-9,16,共3页Journal of Tianjin Normal University：Natural Science Edition

摘　　要：采用误差反传前向人工神经网络模型研究了类黄酮类天然抗氧化剂的结构与其抗氧化能力的关系,用8×4×1网络预测其抗氧化能力,预测误差均小于1%,取得了较好的预测结果,网络自相容能力和交叉检验结果良好.The relationship between structure of flavonoid natural antioxidant and its antioxidant abilities was studied by using neural network based on the back propagation algorithm. The prediction results were examined by a selfconsistency test and a cross-validation test. The self-consistency test and the cross-validation test obtained good results when using the physical property parameters about structure as the inputs, their errors were lower 1%.

关 键 词：BP网络 类黄酮 构性关系 

分 类 号：O641.12[理学—物理化学]