基于BP神经网络的毛脉酸模根中白藜芦醇的动态规律研究  

Study on Accumulative Dynamic Variation of Resveratrol in Root of Rumex gmelini Based on Back Propagation Neural Network

作　　者：王宗权[1,2] 王振月[2] 崔红花[2,3] 康毅华[2] 

机构地区：[1]河北以岭医药集团医药研究院,河北石家庄050035 [2]黑龙江中医药大学药学院,黑龙江哈尔滨150040 [3]广州中医药大学中药学院,广东广州510405

出　　处：《时珍国医国药》2008年第11期2582-2584,共3页Lishizhen Medicine and Materia Medica Research

基　　金：国家自然科学基金(No.30270156)

摘　　要：目的利用BP神经网络方法探索建立毛脉酸模根中白藜芦醇含量的动态模型的可行性。方法采用HPLC色谱法,测定不同生长发育期毛脉酸模中7种生物活性成分(白藜芦醇苷、白藜芦醇、大黄酚苷、酸模素、大黄素、大黄酚、大黄素甲醚)的含量。通过相关性分析找出与白藜芦醇相关性较好的因子用来建立网络。结果通过利用根皮部的白藜芦醇的数据进行检验网络的泛化性能,发现网络的输出值与实际值吻合度较好。结论BP人工神经网络预测白藜芦醇的动态规律比其他线性数值模拟预测具有较大的优势,为植物的生物活性成分随季节的变化趋势的建立提供了一种新的方法。Objective To study the feasibility of dynamic model of resveratol in root of Rumex gmelini by using Back propagation neural network. Methods Seven active component （ Polydatin, Resveratrol, Chrysophan, Rumicin, Emodin, Chrysophanol, Physcion ） contents of R. gmelini which were collected at different developing stages were measured by HPLC, and for the most influential elements which could reflect the trends of aquatic ecology were looked for. Results We used the data of resveratrol in root bark from the universality of the network, and found the outputs tallied in accordance with the measured values very well. Conclusion Artificial neural network is an effective method for forecasting accumulative dynamic variation of resveratrol, has greater advantage than other linearity numeric modeling, and can supply original method for erecting active component in plants to follow changed tendency.

关 键 词：毛脉酸模 BP人工神经网络 白藜芦醇 

分 类 号：R284.1[医药卫生—中药学]