人工神经网络在虎杖提取过程质量控制的研究  

Study on Fast Quality Control in Extracting Process of Rhizoma Polygoni Cuspidati with ANN

作　　者：汪学楷[1] 杨阳[1] 武佳[1] 李晖[1] 

机构地区：[1]四川大学化学工程学院,四川成都610065

出　　处：《四川大学学报（工程科学版）》2012年第4期176-179,共4页Journal of Sichuan University (Engineering Science Edition)

摘　　要：针对人工神经网络收敛速度慢及容易陷入局部最优解的缺点,结合遗传算法全局搜索的特点,提出了一种基于遗传算法和人工神经网络建立近红外透射光谱快速测定虎杖提取液中虎杖苷和白藜芦醇含量的方法。结果表明,模型效果良好,方法简便、准确可靠。与偏最小二乘法(PLS)相比较,该方法具有更高的预测能力,适用性更强,有望应用于药物提取过程的中间控制。In view of the shortcomings of artificial neural network, such as slow convergence, easily falling into local optimum, based on ANN and GA, a method was put forward to establish a model which could achieve fast analysis of Polydatin and Resveratrol in the extracting process of Rbizoma Polygoni Cuspidati by near infrared spectroscopy（NIR） . The method was proved to be convenient, rap- id, accurate and reliable, It has the better precision of prediction and applicability compared with PLS, and provides a promising appli- cation for pharmaceutical production process and quality control.

关 键 词：虎杖苷 白藜芦醇 人工神经网络 遗传算法 偏最小二乘法 

分 类 号：O657.33[理学—分析化学]