著名跨国公司的6_Sigma_BB黑带培训资料-__Booster_Wk2_03_DOE1_0704_V3.pptVIP

Introduction To Design Of Experiments Design of Experiments Design of Experiments is an approach for effectively and efficiently exploring the cause-and-effect relationship between numerous process variables (Xs) and the output or process performance variable (Y). Identifies the “vital few” sources of variation (Xs) Those that have the biggest impact on results Quantifies the effects of the important Xs, including their interactions Produces an equation that quantifies the relationship between the Xs and Y You can predict how much gain or loss will result from changes in process conditions Approaches to Root Cause Analysis Observe the process Observe the process “as is” using historical data or special studies Time plots, control charts, stratification Correlation studies using regression analysis Experiment with the process Change the process in a planned way and measure the results Use Design of Experiments (for more than 1 factor) “To determine what happens with a process when you interfere with it, you have to interfere with it, not passively observe it.” —George E. P. Box Using Historical Data—Some Limitations Limitations of existing data Many organizations already have data on a process and would like to use it to understand the process and make improvements. This approach has limitations because: Existing data often contains errors Records are frequently incomplete Missing values Omitted variables (Xs) Important variables may not have varied during the time when the data were collected Process variables could be correlated with other variables—leading to a false impression about their effect on the process Experimenting With the Process Summary We need a better approach to experimentation, an approach that… Produces results we are confident in Allows us to detect real differences even knowing that common cause variation will always be present Distinguishes between important/unimportant factors Finds interactions between factors Quantifies the effec