The Theory Analysis And Practical Application On Multivariate Control Chart

With the development of the VLSI(VLSI: Very Large-scale Integration), a ICs chip could hold more than 10 9transistors, and the failure rate has been reduced to 10 FET level. The requirement has emerged for the new process quality and reliability evaluative technique in the microcircuit manufacture. A multivariate control module is introduced in this paper. Frequently the pattern of points on the multivariate control charts will contain information, about whether the process is under control. It is shown in this paper that with drawing the Hotelling T~2 control charts and the control ellipse at the same time, we can judge whether the process is under control in these two different ways: variable data and the intuitive graphic way. It is also shown in this paper that the interpretation of an out-of-control signal from a T~2 statistic is greatly aided if the corresponding value is partitioned into independent parts.The result from this decomposition could give the information for further analyzing the out-of-control signal. Using the method of decomposition of statistic T~2, with the data analysis of the out-of-control signals, we can finally find out which made the process out of control, univariate variable or the correlation between variables.