|
Taguchi method for parameter design , design - based approach to quality management , its core essence Taguchi method is a non-linear relationship between the use of parameters and parameters with a variety of environmental factors , to improve interaction between the interfering factors anti-interference ability of the product ( system ) , in the manufacture and use of robustness. Factor selection is based on experience in previous applications , the parameter design test is based on , but the lack of a theoretical basis . This study attempts to proceed from a statistical point of view , to establish a set of methods to test factor is divided into high , medium and low level , to guide Taguchi experimental design . First, in-depth analysis of Taguchi , find out the common factor of all tests to examine , from the four aspects of the error in the degree of influence the level of interactivity , easy to control the degree of test to establish the factor evaluation of scientific evaluation of these factors may be the evaluation index; Secondly, based on the evaluation index , the first cluster analysis factor classification scheduled factor Level Next , on the basis of classification , by analysis of variance factor divided into high three levels ; Finally, through the three group compared the effectiveness of the test empirical factor grading method , and find the open low high school three factor hierarchical general discriminant function discriminant analysis , Taguchi experimental design step and proposed a classification based on factor promoting factor classification method . The results of this study show that the effective guidance Taguchi experimental design factor factor classification can choose to improve the Taguchi test the efficiency , the purpose of achieving a higher degree of optimization under the same experimental conditions . I.e. , in the experimental design , the test factor selected priority select test factor Orthogonal test, to make a higher test under the same test number of factors , the conditions of the same number of tests , the degree of optimization from the advanced factor - smaller fluctuations , placement of more focused, smaller mass loss .
|