Lecture_November_3

Lecture_November_3 - 1 Stat Stat 430/830 430/830 Analysis...

This preview shows pages 1–4. Sign up to view the full content.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 1 Stat Stat 430/830: 430/830: Analysis Analysis of of Variance Variance Lecture November 3, 2010 Response Surface Methods (RSM) Response Surface Methods (RSM) and Designs and Designs Techniques used to optimize the response influenced by several controllable variables. Example: suppose we need to optimize the yield in a process influenced by temperature ( x 1 ) and pressure ( x 2 ). The process yield can be represented as y = η (x 1 , x 2 ) + ε where ε represents a noise or error in the observed response. The function η (x 1 , x 2 ) = E(y) is called the response surface RSM RSM Contour Plot Response Surface RSM RSM Usually we do not know the form of η . Then we approximate it using low-order polynomial responses. First-order (linear) model y = β o + β 1 ⋅ x 1 + … + β k ⋅ x k + ε Second-order (quadratic) model ε β β β β + + + + = ∑ ∑ ∑ ∑ < = = j i j i ij k i i ii k i i i o x x x x y 1 2 1 2 RSM RSM – Bias and Variance Bias and Variance bias. x x x x x x x x x or error c systemati the is ) f(- ) ( ) ( where ) ( ) ( f ) ( y Then, ). R( interest of region the over ) ( f ~ E(y) e approximat we Assume . x (x let and variables coded the are ( 0.5(- x Assume k 1 i i i i i i η δ ε δ ε η ξ ξ ξ ξ ξ = + + = + = =- + =- +- + ) , ... , ) ) RSM RSM – Bias and Variance Bias and Variance ( ) ( ) ( ) ( ) ( ) ( ) 2 2 2 1 2 2 2 2 2 2 2 2 2 2 2 1 ˆ . / ) ( 1 ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ x x N i i x x 1- x V )/ y NV( variance normalized The N x x with x N 1 x)' (1 X) (X' x) (1 bx) V(a ) y V( E(b)x E(a) ) y E( Then bx. a y equation square least the provides x with points data N of set particular a Suppose . ) (- ) y E( y V ) (- ) y E( ) y ( E- y E ) (- ) y E( ) y E(- y E ) (- y E is ) f( y ion approximat the of error square mean normalized The σ σ σ σ σ η η η η + = =- = + = ⋅ ⋅ = + = + = + = = + = + = + = = ∑ = Example x x x x x RSM RSM – Bias and Variance Bias and Variance 32 27 1 3 / 4 8 27 1 3 / 2 5 . 1 1 2 2 2 2 x V x : (b) Design x V x : (a) Design x x V Then ). x ,0, (-x at points three just with designs two Consider . x o x o o x o o + = = + = = + = Example RMS Sequential Procedure RMS Sequential Procedure First Order Model First Order Model Objective Find a path of improvement towards the vicinity of the optimum 3 Steepest Ascent / Descent Steepest Ascent / Descent First Order Model First Order Model Steepest Ascent : move in the direction of maximum increase. Maximization is desired Steepest Descent : move in the direction of maximum decrease. Minimization is desired For a linear model the contours are parallel lines. Steps proportional to the regression coefficients β i ....
View Full Document

{[ snackBarMessage ]}

Page1 / 14

Lecture_November_3 - 1 Stat Stat 430/830 430/830 Analysis...

This preview shows document pages 1 - 4. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online