Measurement System Analysis For Correlated Profile And Multivariate Quality Characteristic

As a modern quality tool for quality control and improvement in Six Sigma management,measurement system analysis is used more and more frequently.The purpose of measurement system analysis is to judge the adequacy of measurement system capability by analyzing the source of fluctuation and variation of measurement system and its impact,and finally to determine whether the measurement system is acceptable.In actual production,the quality of production products or processes is usually detected or guaranteed by means or tools of statistical quality control.And statistical quality control is based on measurement.The ability of measurement system must be evaluated and judged to ensure the reliability and effectiveness of measurement data.Generally,the quality characteristics of the products or processes could be divided into univariate,bivariate or multivariate according to their complexity.For the manufacturing process of complex and precise products,the key quality characteristics of products were often described by profile models,and there were even situations of correlated profile and multivariate quality characteristics.Because of the complexity of profile data and the correlation between profile and multivariate quality characteristics,it was more difficult to evaluate capability for this quality characteristic measurement system.However,the traditional univariate or multivariate capability evaluation methods could not be directly applied to the analysis of the profile data measurement system.Therefore,how to analysis the measurement system for correlated profile and multivariate quality characteristics is a problem that remains to be solved at present.Some scholars at home and abroad have studied the process capability analysis or control charts of simple linear profile,but there is a lack of research on measurement system analysis for correlated profile and multivariate quality characteristics.Because of these problems,the method of constructing a combination of multidimensional statistical vectors is used to analyze and evaluate capability of the measurement system for correlated profile and multivariate quality characteristics.Firstly,two parameters of simple linear profile are composed into two-dimensional predictive profile vectors.Then,according to their correlation between them,two-dimensional profile vectors and multivariate are combined into a new multidimensional statistical vector.The problem of profile and multivariate is transformed into the problem of multivariate problem.Next,the dimension of this statistical vector is reduced by principal component analysis.Then carry out study of the gauge repeatability and reproducibility for simple linear profile and multivariate measurement system and obtain the evaluation indexes of measurement system capability.Finally,the simulation experiment is carried out to realize measurement system analysis and solve the problem of measurement system capability evaluation for simple linear profile and multivariate quality characteristic through the simulation calculation.The results show that the simulation experiment proves the effectiveness and feasibility of the applied method well.