This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: ME 530.343: Design and Analysis of Dynamic Systems Spring 2009 Lecture 3 – Developing Models From Data Monday March 2, 2009 1 Today’s Objectives • Linearization examples • Function identification and parameter estimation • Fitting models to scattered data 2 Linearization Examples 2.1 Classic Pendulum Example ¨ θ + g L sin( θ ) = 0 = ⇒ f ( θ, ¨ θ ) = ¨ θ + g L sin( θ ) = 0 We want to linearize about the reference angle, θ r . ¨ θ r = 0 (since θ r is a constant) Now, write out Taylor series expansion to the first derivative. Note that this is a multivariable expansion in θ and ¨ θ r : f ( θ, ¨ θ ) ≈ f ( θ r , ¨ θ r ) + ∂f ∂θ θ r , ¨ θ r ( θ θ r ) + ∂f ∂ ¨ θ θ r , ¨ θ r ( ¨ θ ¨ θ r ) f ( θ, ¨ θ ) ≈ f ( θ r , 0) + ∂f ∂θ θ r , ( θ θ r ) + ∂f ∂ ¨ θ θ r , ( ¨ θ ¨ θ r ) 1 Now we evaluate terms on right hand side: f ( θ, ¨ θ ) = ¨ θ + g L sin( θ ) f ( θ r , 0) = g L sin( θ r ) ∂f ∂θ θ r , = g L cos( θ r ) ∂f ∂ ¨ θ θ r , = 1 Thus, f ( θ, ¨ θ ) ≈ g L sin( θ r ) + g L cos( θ r )( θ θ r ) + 1( ¨ θ ¨ θ r ) To have a valid solution, the first term must equal zero (recall that f ( θ r , ¨ θ r ) = 0) g L sin( θ r ) = 0 → θ r = nπ , n ∈ Z (all multiples of π ). Hence f ( θ, ¨ θ ) ≈ g L cos( θ r )( θ θ r ) + ( ¨ θ ¨ θ r ) where cos θ r = +1 pendulum is “down” ( n even) cos θ r = 1 pendulum is “up” ( n odd) Now we will consider stable and unstable solutions to the pendulum problem 2.2 Stable solutions When θ r = 0 , ± 2 π, ± 4 π, . . . . We assume that θ r = 0, since these are all equivalent physically.= 0, since these are all equivalent physically....
View
Full
Document
This note was uploaded on 03/02/2010 for the course MECH 530.343 taught by Professor Sun during the Spring '08 term at Johns Hopkins.
 Spring '08
 Sun

Click to edit the document details