|
|
|
An experiment can have various objectives: to quantify the value of a variable or the range in which a component can be used, improve a process or the performance of a machine. To execute a test is usually expensive, therefore test optimization is useful to grant a maximum of information with the minimum amount of trials: an attentive management of tests and the knowledge of mathematical and statistical techniques allow us to extrapolate information that otherwise would stay hidden.
One of the objectives of a serie of tests is to identify the parameters characterizing a model of the real phenomenon that we are studing. For exemple, direct proportionality between variables uses a linear equation and it is identified with a linear regression.
The specialist of the design of experiments ( D.o.E.) uses the variables that the specialist of the process gives him and defines the way to do the trials. In fact, a good way to execute the test allows to minimize the casual error, also called measurement noise.
|
|
