Quality Testing Of Chinese Medicinal Materials Based On Machine Olfaction

At present our country’s medicine market is confusion relatively. The phenomenon of fake medicine instead of true is quite serious that appearance is similar, but drug efficacy different. It has affected the synthesis drugs medical effect, and harmed patient’s benefit enormously, in addition has the possibility to harm patient’s life. Therefore, it is very important to examine medicine’s smell accurately in the traditional Chinese medicine circulates and stores, discover shoddy goods promptly, control traditional Chinese medicine quality and reduce the possibility to enter the drugs market. In order to provide the science basis for attacking false and disqualification product, maintain our country’s medicine market, protect consumer’s legitimate rights and interests.To establish traditional Chinese medicine performance test standard system which has our country’s independent intellectual property rights, is also the crucial question that must solve to achieve the traditional Chinese medicine standardization, the modernization, internationalization. Based on the detailed understanding and analysis of the traditional Chinese medicine classification distinction and the quality detection aspect’s domestic and foreign present situations, this paper has introduced one new technology--the machine sense of smell technology, which distinguish and evaluate category of traditional Chinese medicine through the smell synthesis information.This topic’s main research mentality is:The traditional Chinese medicine as the research object, and the electronic nose as examination method. This paper achieve the different types of traditional Chinese medicine volatile constituent smell fingerprint using the machine sense of smell technology, and analysis the sensor response signal, which takes each sensor’s biggest response value as primitive character parameter vector, to complete the characteristic production; Then using the principal components analytic method, to discovers which can most represent the characteristic parameter of overall information, realize the data dimensionality reduction and the feature extraction; Finally realizes traditional Chinese medicine type distinction through support vector machinesThis paper first introduces the identification of traditional Chinese medicinal materials research, as well as the present situation and research progress in food detection of the machine olfaction, and then this paper analyzes support vector machine used in pattern recognition. Then on the basis of these theories, the support vector machine is used for classification of traditional Chinese medicine. In parameter optimization, not only the common cross validation algorithm is used, but also particle swarm optimization algorithm is used, and particle swarm optimization algorithm for support vector machine obtain good result. Compared with cross validation optimization, particle swarm optimization optimize support vector machine is better. In addition, neural network is used for classification of traditional Chinese medicine but test classification of support vector machine is prior to neural network. So support vector machine can use for classification of traditional Chinese medicine.