爱立信的doe英文版(编辑修改稿)

爱立信的doe英文版(编辑修改稿)内容摘要：

variables and their values • Design the experimental array select the best (least trials) orthogonal array(s) for independent variables assign variables (and interactions) to array columns • Plan the experiment establish plans for process stability and taking measurement answer questions。 what, when, who, where, and how prepare randomisation and repetition plans determine controls for „constant‟ factors Planning for success key steps EN/FAD 109 0015 169。 Ola Johansson, 1999 Ericsson Quality Management Institute Design of Experiments Common Mistakes  Setting levels Too close means no difference in response will be detected Too wide means nonlinearities might pass undetected  Considering Noise Failing to recognise and deal with nuisance variables  Emphasis On Curvature Belief in plexity rather than simplicity  Desire To Optimise Need to isolate “vital few” and identify key leverage variables before optimising  Randomise Trials Using standard run order leading to noise impact trends and bias  Ignoring Interactions Leading to pounding/aliasing and sub optimal results  Confirmation Run Checking results of experiment before implementation  Planning / Preparation Insufficient time given to deciding on design plan, logistics and quality characteristics (may be several) EN/FAD 109 0015 169。 Ola Johansson, 1999 Ericsson Quality Management Institute DOE Why Industrial Experiments Fail • Attacking one variable / response at a time. • Ignoring the possibility of interactions • Failing to recognise and deal with nuisance variables • Putting too much emphasis on curvature (belief in plexity rather than simplicity) • Designing to optimise the process before establishing important variables • Not sequencing trials to randomise time effects • Believing you are the seat of all wisdom and excluding others • Too much fiddling with the levels of key variables • Throwing out observations on the excuse they are outliers • Failing to carry out a confirmation run EN/FAD 109 0015 169。 Ola Johansson, 1999 Ericsson Quality Management Institute A Two Level Factorial Design  Effect of pressure is difference in mean response between runs where it is at plus (high) level and minus (low) level  Effect is units but do not know if this is significant  Analysing the design calculating the “effect” of changing the factors : level minus (low) to level plus (high) Std Order 1 2 3 4 5 6 7 8 Run Order 4 7 3 1 5 8 6 2 Pressure 55 70+ 55 70+ 55 70+ 55 70+ Heat Med Med High + High + Med Med High + High + Time 6 6 6 6 8+ 8+ 8+ 8+ Response Effect of pressure = + + + + + + = 4 4 To pare the mean values where pressure is high and low EN/FAD 109 0015 169。 Ola Johansson, 1999 Ericsson Quality Management Institute A Two Level Factorial Design • Analysing the design use the contrast method • Add the responses, taking account of minus/plus signs (modulus) and then divide by number of runs divided by 2 Pressure + + + + Heat + + + + Time + + + + Pr*He + + + + Pr*Ti + + + + He*Ti + +。