Control Systems. Dynamic response in the time domain. L. Lanari

Similar documents
Control Systems. Internal Stability - LTI systems. L. Lanari

Control Systems. Time response

Control Systems. Laplace domain analysis

Control Systems. Internal Stability - LTI systems. L. Lanari

Control Systems. Time response. L. Lanari

Control Systems. Linear Algebra topics. L. Lanari

The goal of this chapter is to study linear systems of ordinary differential equations: dt,..., dx ) T

Control Systems. Frequency domain analysis. L. Lanari

Examples include: (a) the Lorenz system for climate and weather modeling (b) the Hodgkin-Huxley system for neuron modeling

21 Linear State-Space Representations

Jordan normal form notes (version date: 11/21/07)

Section 9.8 Higher Order Linear Equations

systems of linear di erential If the homogeneous linear di erential system is diagonalizable,

MTH 464: Computational Linear Algebra

Remark 1 By definition, an eigenvector must be a nonzero vector, but eigenvalue could be zero.

Remark By definition, an eigenvector must be a nonzero vector, but eigenvalue could be zero.

PRACTICE FINAL EXAM. why. If they are dependent, exhibit a linear dependence relation among them.

Understanding the Matrix Exponential

Definition (T -invariant subspace) Example. Example

Exercise Set 7.2. Skills

Diagonalization. P. Danziger. u B = A 1. B u S.

Eigenvalues and Eigenvectors

Diagonalization of Matrix

4. Linear transformations as a vector space 17

Topic # /31 Feedback Control Systems

Even-Numbered Homework Solutions

Homogeneous Linear Systems of Differential Equations with Constant Coefficients

Differential equations

Chapter 7. Canonical Forms. 7.1 Eigenvalues and Eigenvectors

State will have dimension 5. One possible choice is given by y and its derivatives up to y (4)

Generalized Eigenvectors and Jordan Form

Lecture Notes for Math 524

Solution of Linear State-space Systems

Identification Methods for Structural Systems

ELEC 3035, Lecture 3: Autonomous systems Ivan Markovsky

1. Find the solution of the following uncontrolled linear system. 2 α 1 1

1 Linear Algebra Problems

Eigenvectors. Prop-Defn

Linear Algebra, part 2 Eigenvalues, eigenvectors and least squares solutions

APPPHYS 217 Tuesday 6 April 2010

Module 09 From s-domain to time-domain From ODEs, TFs to State-Space Modern Control

Lecture 7. Please note. Additional tutorial. Please note that there is no lecture on Tuesday, 15 November 2011.

Eigenvalues and Eigenvectors

Jordan Canonical Form Homework Solutions

Example: Filter output power. maximization. Definition. Eigenvalues, eigenvectors and similarity. Example: Stability of linear systems.

Linear ODEs. Existence of solutions to linear IVPs. Resolvent matrix. Autonomous linear systems

Reduction to the associated homogeneous system via a particular solution

2.3. VECTOR SPACES 25

Matrix Theory. A.Holst, V.Ufnarovski

AMS526: Numerical Analysis I (Numerical Linear Algebra for Computational and Data Sciences)

DIAGONALIZATION. In order to see the implications of this definition, let us consider the following example Example 1. Consider the matrix

Synopsis of Numerical Linear Algebra

x 3y 2z = 6 1.2) 2x 4y 3z = 8 3x + 6y + 8z = 5 x + 3y 2z + 5t = 4 1.5) 2x + 8y z + 9t = 9 3x + 5y 12z + 17t = 7

In these chapter 2A notes write vectors in boldface to reduce the ambiguity of the notation.

Module 03 Linear Systems Theory: Necessary Background

Final Exam Practice Problems Answers Math 24 Winter 2012

CHAPTER 3. Matrix Eigenvalue Problems

Course 216: Ordinary Differential Equations

Linear Algebra Practice Problems

Linear Systems Notes for CDS-140a

Eigenvalues, Eigenvectors. Eigenvalues and eigenvector will be fundamentally related to the nature of the solutions of state space systems.

DIAGONALIZABLE LINEAR SYSTEMS AND STABILITY. 1. Algebraic facts. We first recall two descriptions of matrix multiplication.

Section 3: Complex numbers

SYSTEMTEORI - ÖVNING 1. In this exercise, we will learn how to solve the following linear differential equation:

Lecture 6. Eigen-analysis

6.207/14.15: Networks Lectures 4, 5 & 6: Linear Dynamics, Markov Chains, Centralities

22.2. Applications of Eigenvalues and Eigenvectors. Introduction. Prerequisites. Learning Outcomes

Maths for Signals and Systems Linear Algebra in Engineering

Systems of Linear ODEs

Diagonalizing Matrices

Contents. Preface for the Instructor. Preface for the Student. xvii. Acknowledgments. 1 Vector Spaces 1 1.A R n and C n 2

The Exponential of a Matrix

7 Planar systems of linear ODE

22m:033 Notes: 7.1 Diagonalization of Symmetric Matrices

c Igor Zelenko, Fall

Math 304 Answers to Selected Problems

Linear Algebra- Final Exam Review

Linear Algebra 1. M.T.Nair Department of Mathematics, IIT Madras. and in that case x is called an eigenvector of T corresponding to the eigenvalue λ.

October 4, 2017 EIGENVALUES AND EIGENVECTORS. APPLICATIONS

REPRESENTATION THEORY WEEK 7

Jordan Canonical Form

Autonomous system = system without inputs

1 Systems of Differential Equations

(b) If a multiple of one row of A is added to another row to produce B then det(b) =det(a).

EE263: Introduction to Linear Dynamical Systems Review Session 6

Math Ordinary Differential Equations

Problem Set (T) If A is an m n matrix, B is an n p matrix and D is a p s matrix, then show

September 26, 2017 EIGENVALUES AND EIGENVECTORS. APPLICATIONS

Control Systems I. Lecture 4: Diagonalization, Modal Analysis, Intro to Feedback. Readings: Emilio Frazzoli

MATH 53H - Solutions to Problem Set III

MIT Final Exam Solutions, Spring 2017

EIGENVALUES AND EIGENVECTORS 3

A = 3 1. We conclude that the algebraic multiplicity of the eigenvalues are both one, that is,

Multiple Degree of Freedom Systems. The Millennium bridge required many degrees of freedom to model and design with.

spring, math 204 (mitchell) list of theorems 1 Linear Systems Linear Transformations Matrix Algebra

LINEAR ALGEBRA 1, 2012-I PARTIAL EXAM 3 SOLUTIONS TO PRACTICE PROBLEMS

ELEMENTARY LINEAR ALGEBRA

+ i. cos(t) + 2 sin(t) + c 2.

Solutions Chapter 9. u. (c) u(t) = 1 e t + c 2 e 3 t! c 1 e t 3c 2 e 3 t. (v) (a) u(t) = c 1 e t cos 3t + c 2 e t sin 3t. (b) du

Eigenvalues and Eigenvectors

Transcription:

Control Systems Dynamic response in the time domain L. Lanari

outline A diagonalizable - real eigenvalues (aperiodic natural modes) - complex conjugate eigenvalues (pseudoperiodic natural modes) - phase plots A not diagonalizable - Jordan blocks and corresponding natural modes - special case: Re( i) = Lanari: CS - Dynamic response - Time

what we know S ( ẋ(t) = Ax(t)+Bu(t) y(t) = Cx(t)+Du(t) x() = x x R n u R p y R q A : n n B : n p C : q n D : q p u input ẋ = Ax + Bu y = Cx + Du state y output solution x(t) = (t)x + y(t) = (t)x + Z t Z t H(t )u( )d W (t )u( )d with (t) =e At (t) =Ce At H(t) =e At B W (t) =Ce At B + D (t) Lanari: CS - Dynamic response - Time

what we need we want to analyze the general solution so to qualitatively describe the motion of our system and understand some of its basic properties to go further we need to be able to easily compute the exponential e At = X k= t k k! Ak which appears in all 4 terms and thus contributions (t) =e At (t) =Ce At H(t) =e At B W (t) =Ce At B + D (t) Lanari: CS - Dynamic response - Time 4

how can linear algebra help? ẋ = Ax + Bu y = Cx + Du if 9 T : z = Tx det(t ) 6= same system - different state ż = e Az + e Bu y = e Cz + e Du with e e At easier to compute e At = T e e At T then e e At = e TAT t = Te At T e e At easier to compute if is also easier to compute what special structure needs to have matrix ea in order to make easier to compute? e e At Lanari: CS - Dynamic response - Time 5

easiest case = diag{ i } = 6 4... n 7 5 k = diag{ k i } = 6 4 k k... 7 5 k n e t = X k= k tk k! = = 6 4 P k= k t k /k! P k= k t k /k!... P k= k nt k /k! 7 5 e t = 6 4 e t e t... 7 5 = diag{e it } e nt matrix exponential of a diagonal matrix is immediate Lanari: CS - Dynamic response - Time 6

how can linear algebra help? diagonalizable case if the matrix is diagonal then its matrix exponential is immediate we found that a square matrix A could be diagonalized iff mg( i) = ma( i) for all i with the diagonalizing change of coordinates T defined as T = U e At = T e t T = U e t U Lanari: CS - Dynamic response - Time easy to compute 7

how can linear algebra help? non-diagonalizable case if ea =diag{j i } is block diagonal is the matrix exponential also simplified? e diag{j i}t = X k= diag{j i } k tk k! = = 6 4 e J t e J t yes... 7 5 = diag{ej it } e Jrt moreover diag{e J it } has a special structure that we are going to explore (being Ji a Jordan block) Lanari: CS - Dynamic response - Time 8

summary A diagonalizable e At = U e t U e At A non-diagonalizable e At = T diag e J it T what s next? we are going to explore these two cases and understand the different time functions that are present so that we know how, for example, the ZIR qualitatively behaves Lanari: CS - Dynamic response - Time 9

matrix exponential: A diagonalizable e At = U e t U T = U = u u u n e At = u u... u n 6 4 e t e t... 7 5 e nt v T v T 6 4. 7 5 v T n spectral form of the matrix exponential e At = nx i= e it u i v T i if A diagonalizable (also true for complex pairs of eigenvalues) Lanari: CS - Dynamic response - Time

( matrix exponential: ZIR (if A diagonalizable) x ZIR (t) = nx e it u i vi T x natural modes i= only function of time projection of the initial condition on the subspace generated by ui constant How the zero-input response (unforced response) varies in time depends upon the eigenvalues qualitative behavior of the system motion i real complex Lanari: CS - Dynamic response - Time

matrix exponential: A diagonalizable we distinguish between real and complex eigenvalues if A diagonalizable if real eigenvalue natural modes e i t if complex eigenvalues ( i, i = i ) we need to expand this i + j i apple e it e i t e or 4 i i 5t i i preferred Lanari: CS - Dynamic response - Time

A diagonalizable - real eigenvalues if real eigenvalue e i t i > aperiodic mode i = /e i i < t e it = e t/ i with i = i time constant (meaningful only for negative i) Lanari: CS - Dynamic response - Time

A diagonalizable - real eigenvalues example: real eigenvalue (n = ) x ZIR (t) = > < nx i= e it u i v T i x for n = x ZIR (t) =e t u v T x + e t u v T x projection matrices = e t u c + e t u c scalars eigenspace < x > eigenspace x u u aperiodic modes x x x x x Lanari: CS - Dynamic response - Time 4

initial conditions x ZIR (t) = nx i= e it u i v T i x how to see the contribution of an initial condition to each natural mode x ( u i vi T use the projection matrices = P i u i vi T x = P i x = c i u i or x = nx i= c i u i express the initial condition in the base given by the eigenvectors Lanari: CS - Dynamic response - Time 5

examples v = m F A = apple Mass-Spring-Damper (MSD) m s + µṡ + ks = u A = apple k m µ m real eigenvalues when high damping µ p km compute eigenvalues and check, for the real case, the sign discuss how the eigenvalues and therefore the ZIR varies with µ in the real eigenvalues case Lanari: CS - Dynamic response - Time 6

example: chemical reaction first order chemical reactions reversible reaction C A C B concentration component A concentration component B A k d! B B k i! A dc A dt dc B dt = k d C A + k i C B = k d C A k i C B (find eigenvalues, interpret) Ċ A + ĊB = mass conservation C A (t)+c B (t) =C A () + C B () Lanari: CS - Dynamic response - Time 7

example: chemical reaction A k! B k! C dc A dt dc B dt dc C dt = k C A = k C A k C B = k C B (find eigenvalues, interpret) C A (t)+c B (t)+c C (t) = initial value Lanari: CS - Dynamic response - Time 8

example: chemical reaction parallel reaction A k! B A k! C A k! D dc A dt dc B dt dc C dt dc D dt = (k + k + k ) C A = k C A = k C A = k C A Lanari: CS - Dynamic response - Time 9

A diagonalizable - complex eigenvalues if A diagonalizable and i = i + j!i the free state response is (ZIR) we need to: 9 T R such that T R AT R = apple x ZIR (t) =e At x o = T R i i i i 4 i! i e! i i (real form) 5t T R x compute e 4 i i 5t i i find the change of coordinates TR that puts a generic ( x ) matrix A with complex eigenvalues into the real form apple T R AT R = i i i i Lanari: CS - Dynamic response - Time

A diagonalizable - complex eigenvalues ( i, i ) pair with i = i + j i e 4 i i 5t i i = e = e @ 4 i 5t+ i 4 i 5t i = e it I 4 i i 4 i e i X apple i i k= 5tA 5t k t k k!! since matrices commute definition of exponential =... = e i t apple cos i t sin i t sin i t cos i t recognize known series e 4 i i 5t i i = e i t apple cos i t sin i t sin i t cos i t Lanari: CS - Dynamic response - Time

A diagonalizable - complex eigenvalues change of coordinates for the real system representation if complex eigenvalues i = i + j!i eigenvector complex vector u i = u re + ju im real vectors T R = u re u im non-singular also linearly independent write the initial condition as x = c a u re + c b u im = u re apple c u a im = T c R b apple ca c b T R x = apple ca c b define the quantities m R and ' R as m R = q c a + c b sin R = cos R = c a p c a + c b c b p c a + c b c a = m R sin c b = m R cos R R Lanari: CS - Dynamic response - Time

A diagonalizable - complex eigenvalues e At x = T e 4 i i 5t i i Tx = u re u im e it apple cos i t sin i t sin i t cos i t apple mr sin R m R cos R depends upon the initial condition ZIR e At x = m R e it [sin( i t + R )u re + cos( i t + R )u im ] vectors exponential x periodic Lanari: CS - Dynamic response - Time

natural modes (complex - A diagonalizable) if complex eigenvalues ( i, i ) pseudoperiodic mode e At x = m R e it [sin( i t + R )u re + cos( i t + R )u im ] e i t t e t t e i t t i > i = i < time functions of the form e t sin(!t) e t cos(!t) from and! we have a qualitative behaviour of the ZIR Lanari: CS - Dynamic response - Time 4

natural modes (complex - A diagonalizable) pseudoperiodic mode x u im x x u im u re u re i > i = i < (prove how it turns) Lanari: CS - Dynamic response - Time 5

natural modes (complex - A diagonalizable) pseudoperiodic mode i = + j alternative characterization w.r.t. real and imaginary part! n = p +! = p +! natural frequency damping =sin# Im contribution in a characteristic polynomial ( i)( i )= +! n +! n! n #! inverse relations p =! n! =! n p so / =! n ± j! n =! n ± j p Re time functions e! nt sin(! n p t) e! nt cos(! n p t) Lanari: CS - Dynamic response - Time 6

natural modes (complex - A diagonalizable) pseudoperiodic mode i = + j =sin#! n! n! n # Im!! n # Im! Re Re =! n same same! n = study how does the time behavior change when we compare and Lanari: CS - Dynamic response - Time 7

phase plane examples 5 Phase portrait (lambda = -, lambda = ) 4 Phase portrait (lambda = -.5, lambda = -) 4 Phase portrait (lambda = -.5, lambda = ) 4 x x x - - - - - - - -4 - - -5-4 -5-4 - - - 4 5-5 -4 - - - x x -4-5 -4 - - - x = = =.5 = =.5 = 4 Phase portrait (complex conjugate case) 4 Phase portrait (complex conjugate case) 5 Phase portrait (lambda =.5, lambda =.) 4 x x x - - - - - - - - - -4-4 -5-4 - - - x / =.5 ± j -4-5 -4 - - - x / =. ± j -5-5 -4 - - - 4 5 x =.5 =. Lanari: CS - Dynamic response - Time 8

natural modes (Mass - Spring - Damper) pseudoperiodic mode s k u m s + µṡ + ks = u µ A = apple k m µ m eigenvalues real eigenvalues when high damping µ p km if > over damping if = critical damping complex eigenvalues when low damping µ< p km if < under damping n = natural frequency r k m = damping coefficient µ p km mechanical friction coefficient Lanari: CS - Dynamic response - Time 9

natural modes (Mass - Spring - Damper) pseudoperiodic mode m s + µṡ + ks = u Normalized Zero Input Response - Oscillator harmonic oscillator s(t)/s().8.6.4. µ = µ crit µ = µ crit µ =.5µ crit µ = µ crit µ =.7µ crit µ =.µ crit µ = µ crit = p km -. -.4 -.6 -.8-4 5 6 7 8 9 t normalized (s(t)/s()) plot of the ZIR under-damped critically damped over-damped Lanari: CS - Dynamic response - Time

natural modes (diagonalizable case) example: aperiodic + pseudoperiodic mode block-diagonal 4 5 same eigenvalues / = ± j = 4 6 5 4 5 6 65 9.8.6.4..5.5.5.5.5.5..4.6.8.5.5 Lanari: CS - Dynamic response - Time

natural modes (non diagonalizable case) example: one Jordan block of dimension J = 4 i i i 5 = J J 4 i {z } {z } i 5 + 4 5 i diagonal nilpotent commute in the product e (J +J )t = e J t e J t (proof) e Jt = 4 e i t te i t t e i t e i t te i t e i t 5 e it,te it, t e i t new time functions maximum exponent depends on the dimension of the Jordan block Lanari: CS - Dynamic response - Time

natural modes (non diagonalizable case) we have e At = T diag e J it T define nk (dimension of the largest Jordan block associated to i) is defined as index of i new time functions (natural modes) appear e it,te it,..., t n k (n k )! e i t depend on the dimension of the Jordan blocks (nk: index of i) Lanari: CS - Dynamic response - Time

natural modes (non diagonalizable case) what kind of contribution these new terms give? t k k! e i t.9.8.7 Functions t k /k! e t k = 4 5 if i is real negative, exponential wins!.6.5.4... 4 5 6 7 8 9 t example 4 Phase portrait, ma =, mg = (lambda = -.5) A =.5.5 x - - - -4-5 -4 - - - x Lanari: CS - Dynamic response - Time 4

natural modes (non diagonalizable case) if we have complex eigenvalues with geometric multiplicity greater than then the following time functions will also appear e it sin i t, te it sin i t,..., t n k (n k )! e it sin i t Functions t^k/k! e^{ t} \sin t.6.5.4 k = 4 5.......4 4 5 6 7 8 9 t Lanari: CS - Dynamic response - Time 5

natural modes (non diagonalizable case) special cases Re( i) = examples i = 4 5 t diverging t i = j!i j i 6 j i 4 j i j i 7 5 or 6 4! i! i 7! 5 i! i sin!it t sin!it diverging Lanari: CS - Dynamic response - Time 6

natural modes (non diagonalizable case) example x.8.6.4. -. -.4 -.6 -.8 -.5 x -.5 (x, x) plane.8 - Re( i) = i = - -.5 projection on the (x,x ) plane x.5 x = @ A A = e At x = @ A t t / @ t A x only constant mode is selected x(t) = @ A x.6.4. -. -.4 -.6 x = @ A constant and t mode are selected x(t) = + @ t A -.8 - - -.8 -.6 -.4 -...4.6.8 x Lanari: CS - Dynamic response - Time 7

summary A diagonalizable mg( i) = ma( i) for all i real i complex i = i + j!i aperiodic mode e i t pseudoperiodic mode e it [sin( i t + R )u re + cos( i t + R )u im ] e At = nx e it u i v T i spectral form i= A not diagonalizable mg( i) < ma( i) real i complex i = i + j!i...,..., t n k (n k )! e i t t n k (n k )! e it sin i t Lanari: CS - Dynamic response - Time 8

vocabulary English natural mode aperiodic/pseudoperiodic natural mode natural frequency damping coefficient spectral form Italiano modo naturale modo naturale aperiodico/ pseudoperiodico pulsazione naturale coefficiente di smorzamento forma spettrale Lanari: CS - Time response 9