A Study On Application Of MTS In Optimizing And Analyzing Multidimensional System

Mahalanobis-Taguchi System (MYS) is a new method for studyingmultidimensional system, and it can be used to measure the degree of abnormality byconstructing and optimizing the measurement scale. In contrast to traditionalmultivariate methods, such as principal component analysis (PCA), discriminationand classification method, stepwise regression, test of additional information, multipleregression, multivariate process control charts, artificial neural networks (ANN) andso on, MTS shares many advantages;however, because it is a new method, and thestudy on MTS is just started, so MTS still needs to be researched, especially on howto control samples and how to analyze root causes for abnormal points. At the sametime, we need to study and explain some statistical and operational questions aboutMTS.In this thesis, the nature of the multidimensional system is analyzed firstly;thenprimary theory which includes two basis, four steps and four characteristics isintroduced. Secondly, three kinds of MTS methods are compared, and then theiradvantages and limitations are summarized. Thirdly, some statistical and operationalquestions about MTS are studied and explained, which makes MTS more operable.Fourthly, MTS is studied in deeper level from two aspects. On one hand, MultivariateStatistical Process Control is modified and used to control optimizedmultidimensional system;on the other hand, MYT Orthogonal Decomposition ismodified and used to analyze root causes for abnormal points. Finally, the thesisoptimizes and analyzes a hospital viscosity of plasma diagnostic system by MTS, andthe result is very good.