针对制约陶瓷膜精制中药技术的膜污染关键问题,依据现代分离科学基本原理,建立可科学表征中药水提液复杂体系的物理化学特征的标准技术规范,以具有代表性的中药及其复方为实验体系,采集膜过程相关参数,建立膜污染基础数据库,综合运用模式识别、人工智能、支持向量机等方法进行数据挖掘、知识发现.应用中药制剂学、物理化学、计算机化学、化学工程学,跨学科交叉研究中药水提液膜污染的规律,为类似复杂体系的膜污染机理与防治提供了一种全新的研究模式.用测试集进行预测,6种中药水提液膜污染度实际值、拟合值的均方误差为0.6%。所得模型拟合精度高,可满足拟合要求.
Technical standards which can scientifically characterize the complex system of Chinese Herb Water Extractions (CHWE) was established according modern separation science and technology. Nearly 200 kinds of representative herbs or herbal compounds were picked out as experimental objects, physicochemical parameters of filtration process using ceramic membrane were collected and the data - base was built based on these parameters. Data mining and knowledge discovery were carried out by using computing method such as pattern recognition, artificial intelligence and Support Vector Machine (SVC). Knowledge of Chinese medicine pharmaceutics, physical chemistry, analytical chemistry, computer chemistry and chemical engineering was applied synthetically to find the fouling mechanism of CHWE filtration process using ceramic membrane. Results show that mean square error (MSE) of actual values and fitted values is 0.6 %, so a novel route to recognize and prevent the membrane fouling of CHWE filtration process and other natural complex system was developed according to our research.