A New Genetic Clustering Method Of Vinegar Nuclear Magnetic Resonance Metabolomics Data

Vinegar is not only the traditional flavor in China, but also a kind of Chinese medicine. It has such important effects as resistance to fatigue and cancer, adjusting blood pressure and blood fat, promoting appetite. Vinegar, made from sorghum, rice, wheat, etc, is a kind of liquid mainly containing acetic acid. Its chemical composition is very compli-cated. And, due to the different ratio of raw materials, production process, it can be devided into many kinds of different quality. There are many types of vinegar, including Shan-xi mature vinegar, Zhen-jiang balsamic vinegar and Fu-jian monascus rice vinegar, which are the most famous. The difference in vinegar types can lead to different taste, usage and curative effect. People often distinguish vinegar by its color or taste, however, there’s no a set of scientific reasonable method of judgement. Therefore, it has become particularly important to cluster different kinds of vinegar. This article adopted a method of clustering analysis improved by genetic algorithm to solve the clustering problem of vinegar. In instance analysis, we analyzed the mature vinegar, balsamic vinegar and rice vinegar respectively. Based on coding the clustering problem by using floating-point method, we chose the Proportion Selecting operation, Random Crossing operation and Random Mutation operation in Genetic Algorithm and combined the Genetic Algorithm with K-means Clustering Algorithm. The paper constructed an appropriate fitness function and set the proper algorithm parameters through trial and error. It gave full play to global optimization capability of Genetic Algorithm and the locality of K-means clustering algorithm. We used the combined algorithm to carried clustering analysis on different vinegar and compare the final result with that dot from the traditional K-means. The results showed that the experimental data can be devided effectively through the improved clustering algorithm and the improve algorithm is of better efficiency than K-means.This paper includes five chapters:Chapter one briefly introduced the research background of nuclear magnetic reso-nance(NMR) technology, metabonomics and the related knowledge; the relevant know- ledge of genetic algorithm(GA), as well as some commonly used method of clustering analysis; at last, illustrated the main work to be completed in this thesis.Chapter two mainly introduced the principle and procedure of K-means, the basic thoery and terminology of GA; and the feasibility of combining GA with K-means are analyzed in this chapter.Chapter three determined the encoding way of genetic clustering algorithm, con-structed the fitness function, designed the selection, crossover, mutation operation and set termination condition of the evolutionary algorithm.Chapter four mainly solved the classification problem of mature vinegar, rice vinegar and balsamic vinegar NMR metabolomics data and compared the results with that got from K-means. In this chapter, we also analyzed and summarized the experimental results, pointed out the deficiencies and the need to improve in the future.Chapter five analyzed and summarized the classified result. It put forward the dedi-ciency in the classification algorithm improved by GA and places that need improvement.