Introduction to DOE(诺基亚公司 DOE 设计实验 培训 教材).pptVIP

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Executing the Experiment Strategy Define the problem—Set up an accurate catapult Establish the experimental objective—Launch the ball 5 meters with minimal variability Select the output (response) variables—Distance (in meters) Select the input (independent) variables ? Start angle ? Ball type ? Number of bands ? Others ? Stop angle ? Band tension ? Band type Choose the levels for the input variables Only two levels allowed Must be wide enough apart to see the differences Select the experimental design Experimental designs are matrices of -1 and +1 values arranged such that certain mathematical characteristics are maintained Execute the experiment and collect data Analyze data ? Main effects plots ? Interaction plots ? Cube plots Draw statistical conclusions Does the experiment provide useful results? Are there important main effects or interactions? Replicate or validate the experimental results Draw practical solutions Implement solutions * Load the CATAPULT.MTW file from your data disks. The file should look like this: If we had randomized the experiment, the RunOrder column would be in random order and would indicate the order in which we would execute each treatment. In this case, the experiment is in standard order. Graphical Data Analysis Catapult.mtw * Visualizing the Data—The Cube Plot Consider our 23 design on the catapult. N_RubBnd Pin_Pos 0.9 2.1 2.4 5.15 3.35 8.2 4.55 1.5 Strt_Ang Run Pin_ N_Rub Strt_ Meters/Yards Pos Bnd Ang Traveled 1 -1 -1 -1 2.10 2 1 -1 -1 0.90 3 -1 1 -1 3.35 4 1 1 -1 1.50 5 -1 -1 1 5.15 6 1 -1 1 2.40 7 -1 1 1 8.20 8 1 1 1 4.55 A B C Response * 0.9000 2.4000 1.5000 4.5500 2.1000 5.1500 3.3500 8.2000 -1 1 Pin_Pos N_RubBnd Strt_Ang -1 1 -1 1 Cube Plot (Data Means) for Dist Minitab Cube Plot This graph is not generally a useful presentation plot. It is usually used by the exp