基于FCM聚类算法的马田系统研究

Study on MTS Based on FCM Clustering Algorithm

马田系统是一种多元系统定量模式识别方法,是数据分类的有效方法,在很多领域都得到广泛应用。构建正常样本数据的基准空间、筛选出有效的项目、确定阈值等是经典马田系统的重要步骤。文章改进马田系统中筛选有效项目的方法,经典马田系统将正交表和信噪比结合起来筛选有效项目;基于FCM聚类算法的马田系统尝试用FCM聚类算法选择有效项目,正交表的每一行作为一个实验方案,对于每种方案,都利用FCM进行聚类,得到样品分类的正确率。把正确率的信噪比作为筛选有效项目的指标,信噪比越大则选择的有效项目越可信。得到有效的检测项目之后,可以优化马田系统的基准空间,提高样品分类的正确率。

MTS is a kind of quantitative pattern recognition method for multiple systems,and it is an effective method of data classification.MTS has been widely used in many fields.It is an important step for the classic MTS to construct the reference space of the normal sample data,to screen out the effective items and to determine the threshold value.In this paper,we improve the method of screening effective project in MTS,and the classic MTS combines the orthogonal table and the signal to noise ratio to screening effective project;MTS based on FCM clustering algorithm attempts to use the FCM clustering algorithm to select the effective project.Orthogonal table＇s each line as an experimental program,for each program,using FCM for clustering to get the correct rate of classification of samples.The correct rate of the signal to noise ratio as the index of screening effective projects;The greater the signal to noise ratio,the more reliable the valid item is selected.After obtaining the effective project,it can optimize the reference space of MTS,and improve the accuracy of the classification of samples.