支持向量机的中药提取浓度软测量

Soft sensing of Chinese traditional medicine extract concentration based on Support Vector Machine

中药生产过程是一个非常复杂的流程工业工艺过程,提取液浓度是生产过程的重要参数之一,其检测速度及精度一定程度上影响着中药生产质量的高低。目前提取过程中对料液浓度的测量缺乏有效的传感器和检测手段,对浓度的判断一般采用人工观察和离线检测,造成不同批次的药液浓度一致性较差。软测量作为一种新型的过程参数检测技术,为解决复杂工业过程参数的检测问题提供了一条有效的途经。本文针对中药提取工段料液浓度无法在线直接检测的问题,分析了中药生产过程提取工段的工作原理,借助软测量建模的优势,通过测量易获得的辅助变量推导出难测量的主导变量,提出了一种基于最小二乘支持向量机建立提取料液浓度的软测量预测模型,实现了中药浓度的在线测量。实验分析和数据研究表明,此方法学习速度快、跟踪性能好、泛化能力强,可较为快速精确的预测中药提取液浓度,能够满足中药提取过程浓度测量的要求,改变了现有的中药提取液浓度的测量方式,提高了中药生产的自动化水平,为生产合格的中药产品提供了基础。

Chinese Traditional Medicine（CTM） production process is a very complex craft process of process industry and concentration of extract is one of the important parameters of production process. The detection speed and precision have an effect on the level of CTM＇S quality. Current extraction process on the material liquid concentration lack the effective sensors and detection methods. The judgment of concentration is generallly proformed by manual observation and off-line testing, leading to different batches of liquid concentration uniformity. Soft sensing, as a new type of process parameter detection technology, provides an effective way to solve the problem of complex industrial process parameters testing. This paper is aiming at the problem of the concentration which cannot be directly online tested during extract of CTM and analyzed the working principle of CTM＇S extraction. We deduced dominant variable difficult to measure by measuring auxiliary variable which are easily accessible with the advantage of the soft sensor modeling to achieve the online measurement of the CTM＇S concentration and presented a concentration prediction model which was based on support vector machine. Study and data shows that this method which has the high learning speed, good performance for tracking and generalization ability forecast precisely extract concentration of CTM and changed the existing CTM＇S extract concentration measurement,improved the level of automation of the CTM＇S production and provided the basis of producing qualified products.