Performance Measures in Dynamic Parameter Design

Dynamic parameter design introduced by Taguchi (1987) is one of the most important tool in quality engineering. It is also known as parameter design in signal-response systems (Miller and Wu, 1996; Wu and Hamada, 2000, chapter 11). The name suggests that the interest lies in a signal-response relationship rather than a single value of the response. Taguchi’s approach to dynamic parameter design can be described as follows. Let Y be the response and M the signal factor. There exists an ideal relationship between the signal and the response given by M Y I β = . In reality due to the presence of noise factors, deviations occur from this ideal function. Therefore a more realistic statistical model is ε β + = M Y , (1) where 0 ) ( = ε E , , and 2 0 ) ( σ ε = Var ε is a random error caused by the noise factors. The β and are functions of the control factors X. The objective of dynamic parameter design is to choose an X to make (1) as close to the ideal relationship as possible. 2 0 σ