Control Systems. Controllability and Observability (Chapter 6)

Similar documents
Chapter 3.1: Polynomial Functions

Fuzzy Neutrosophic Equivalence Relations

MATH Midterm Examination Victor Matveev October 26, 2016

Intermediate Division Solutions

ME260W Mid-Term Exam Instructor: Xinyu Huang Date: Mar

Sx [ ] = x must yield a

D.S.G. POLLOCK: TOPICS IN TIME-SERIES ANALYSIS STATISTICAL FOURIER ANALYSIS

are specified , are linearly independent Otherwise, they are linearly dependent, and one is expressed by a linear combination of the others

Observer Design with Reduced Measurement Information

Grade 3 Mathematics Course Syllabus Prince George s County Public Schools

UNIVERSITY OF TECHNOLOGY. Department of Mathematics PROBABILITY THEORY, STATISTICS AND OPERATIONS RESEARCH GROUP. Memorandum COSOR 76-10

The Excel FFT Function v1.1 P. T. Debevec February 12, The discrete Fourier transform may be used to identify periodic structures in time ht.

Markov processes and the Kolmogorov equations

Study of Energy Eigenvalues of Three Dimensional. Quantum Wires with Variable Cross Section

ENGI 4421 Central Limit Theorem Page Central Limit Theorem [Navidi, section 4.11; Devore sections ]

Basic Probability/Statistical Theory I

SOLUTION. The reactor thermal output is related to the maximum heat flux in the hot channel by. Z( z ). The position of maximum heat flux ( z max

Modeling Micromixing Effects in a CSTR

Quantum Mechanics for Scientists and Engineers. David Miller

Class #25 Wednesday, April 19, 2018

Ch. 1 Introduction to Estimation 1/15

ELEG3503 Introduction to Digital Signal Processing

Linear time invariant systems

After the completion of this section the student. V.4.2. Power Series Solution. V.4.3. The Method of Frobenius. V.4.4. Taylor Series Solution

Chapter 4 : Laplace Transform

Solutions. Definitions pertaining to solutions

K [f(t)] 2 [ (st) /2 K A GENERALIZED MEIJER TRANSFORMATION. Ku(z) ()x) t -)-I e. K(z) r( + ) () (t 2 I) -1/2 e -zt dt, G. L. N. RAO L.

Explicit and closed formed solution of a differential equation. Closed form: since finite algebraic combination of. converges for x x0

Fourier Series & Fourier Transforms

State Space Representation

Result on the Convergence Behavior of Solutions of Certain System of Third-Order Nonlinear Differential Equations

PROBLEMS AND SOLUTIONS 2

MAT 1275: Introduction to Mathematical Analysis

Math 778S Spectral Graph Theory Handout #3: Eigenvalues of Adjacency Matrix

ANOTHER PROOF FOR FERMAT S LAST THEOREM 1. INTRODUCTION

COMP26120: Introducing Complexity Analysis (2018/19) Lucas Cordeiro

IJISET - International Journal of Innovative Science, Engineering & Technology, Vol. 2 Issue 12, December

Solutions to Midterm II. of the following equation consistent with the boundary condition stated u. y u x y

[1 & α(t & T 1. ' ρ 1

Introduction to Optimization Techniques. How to Solve Equations

MAT 1275: Introduction to Mathematical Analysis

NOTE ON APPELL POLYNOMIALS

THE MEASUREMENT OF THE SPEED OF THE LIGHT

Solutions - Homework # 1

Greedy Algorithms. Kye Halsted. Edited by Chuck Cusack. These notes are based on chapter 17 of [1] and lectures from CSCE423/823, Spring 2001.

Frequency Domain Filtering

Lecture 2 Linear and Time Invariant Systems

MASSACHUSETTS INSTITUTE OF TECHNOLOGY Department of Electrical Engineering and Computer Science. BACKGROUND EXAM September 30, 2004.

Relationships Between Frequency, Capacitance, Inductance and Reactance.

Classification of DT signals

Question1 Multiple choices (circle the most appropriate one):

Principal Component Analysis. Nuno Vasconcelos ECE Department, UCSD

Chapter 5: Properties of Solutions

Super-efficiency Models, Part II

(8) 1f = f. can be viewed as a real vector space where addition is defined by ( a1+ bi

Kernels. Nuno Vasconcelos ECE Department, UCSD

. (7.1.1) This centripetal acceleration is provided by centripetal force. It is directed towards the center of the circle and has a magnitude

Numerical Methods for Ordinary Differential Equations

Formula List for College Algebra Sullivan 10 th ed. DO NOT WRITE ON THIS COPY.

Chapter 2. Periodic points of toral. automorphisms. 2.1 General introduction

CHAPTER 5. Theory and Solution Using Matrix Techniques

Theory of Ordinary Differential Equations. Stability and Bifurcation II. John A. Burns

1) When an object is placed outside the center of curvature of a convex mirror, the image is.

CDS 101: Lecture 5.1 Controllability and State Space Feedback

Goal. Adaptive Finite Element Methods for Non-Stationary Convection-Diffusion Problems. Outline. Differential Equation

4. Optical Resonators

REQUIREMENTS FOR EFFICIENT TRANSISTOR OPERATION I B. Icn. I cp

ELEVENTH YEAR MATHEMATICS

The Law of Total Probability, Bayes Rule, and Random Variables (Oh My!)

Multi-objective Programming Approach for. Fuzzy Linear Programming Problems

Chapter 5. Root Locus Techniques

The Simple Linear Regression Model: Theory

Digital Signal Processing. Homework 2 Solution. Due Monday 4 October Following the method on page 38, the difference equation

Minimization of Sensor Activation in Decentralized Discrete Event Systems

Transfer Function Analysis

REVERSIBLE NON-FLOW PROCESS CONSTANT VOLUME PROCESS (ISOCHORIC PROCESS) In a constant volume process, he working substance is contained in a rigid

MATH : Matrices & Linear Algebra Spring Final Review

ACTIVE FILTERS EXPERIMENT 2 (EXPERIMENTAL)

5.1 Two-Step Conditional Density Estimator

ALE 26. Equilibria for Cell Reactions. What happens to the cell potential as the reaction proceeds over time?

E o and the equilibrium constant, K

Claude Elysée Lobry Université de Nice, Faculté des Sciences, parc Valrose, NICE, France.

Principal Component Analysis

Mass Transfer Chapter 3. Diffusion in Concentrated Solutions

Rank tests and regression rank scores tests in measurement error models

Limit Theorems for Jacobi Ensembles

, which yields. where z1. and z2

x 2 x 3 x b 0, then a, b, c log x 1 log z log x log y 1 logb log a dy 4. dx As tangent is perpendicular to the x axis, slope

An object is placed 20 cm in front of a convex mirror that has a radius of curvature of magnitude 16 cm (note this is not the focal length).

ALGEBRAIC GEOMETRY COURSE NOTES, LECTURE 5: SINGULARITIES.

3.6 Condition number and RGA

School of Mechanical Engineering Purdue University. ME375 Transfer Functions - 1

Lecture 8: October 20, Applications of SVD: least squares approximation

Homology groups of disks with holes

Chapter Summary. Mathematical Induction Strong Induction Recursive Definitions Structural Induction Recursive Algorithms

Discrete-Time Systems, LTI Systems, and Discrete-Time Convolution

Bernoulli Numbers. n(n+1) = n(n+1)(2n+1) = n(n 1) 2

A proposition is a statement that can be either true (T) or false (F), (but not both).

Computing the output response of LTI Systems.

SEQUENTIAL ESTIMATION IN A SUBCLASS OF EXPONENTIAL FAMILY UNDER WEIGHTED SQUARED ERROR LOSS *

Transcription:

6.53 trl Systems trllaility ad Oservaility (hapter 6) Geeral Framewrk i State-Spae pprah Give a LTI system: x x u; y x (*) The system might e ustale r des t meet the required perfrmae spe. Hw a we imprve the situati? The mai apprah: Let u= v- Kx (state feedak) the x x (v - Kx); y x D(v - Kx) ( - K)x v; ( - DK)x - Dv The perfrmae f the system is haged y matrix K. Questis: Is there a matrix K s.t. -K is stale? a eig(-k) e mved t desired latis? These issues are related t the trllaility f (*)

Mai Result : The eigevalues f -K a e mved t ay desired latis iff the system (*) is trllale. ther situati: the state x is t mpletely availale. Oly a liear miati f x e.g. y = x a e measured. Hw a we realie u = v-kx? pssile sluti: uild a server t estimate x ased measuremet f y. Mai result : The server errr (differee etwee the real x ad estimated x) a e made aritrarily small withi aritrarily shrt time perid iff (*) is servale. We will arrive at these lusis i hapter 8. efre that we eed t prepare sme tls ad g thrugh these fudametal prlems: trllaility ad servaility. 3 trllaility: Defiiti sider the system x x u x R ; u R trllaility is a relatiship etwee state ad iput. Defiiti: The system r the pair () is said t e trllale if fr ay iitial state x()=x ad ay fial state x d there exist a fiite time T > ad a iput u(t) t[t] suh that T (T-τ) x(t) e x e u( τ)dτ x T mmet: There may exist differet T ad u that satisfy (). s a result there may e differet trajetries startig frm x ad ed at x d. trllaility des t are aut the differee. p. d () 4

t τ 'τ (tτ) '(tτ) W (t) e 'e dτ e 'e d Equivalet ditis: The fllwig are equivalet ditis fr the pair () t e trllale: ) W (t) is sigular fr every t >. ) W (t) is sigular fr at least e t >. 3) Fr every vr v v e t is t idetially er. 4) The matrix G = [ - ] has full rw rak i.e. (G ) =. 5) The matrix M() = [I ] has full rw rak at all. 6) M() has full rw rak at every eigevalues f. Nte: M() has full rw rak if is t a eigevalue f. We ly eed t hek the rak f M() at eigevalues f. t Nte : Of all the ditis ly 4) ad 6) a e pratially verified. 5 Example: Determie the trllaility fr x x u pprah : G [ ] a ( G) fr all pssile a - a - a ad The system t trllale whatever a ad are. pprah : hek M()=[-I ] at =- M( ) a (M(-)) < fr all pssile a ad Same lusi trllaility 6 3

Example: x x u a pprah : G [ ] a - a - ρ(g ) if a ad detg a ρ(g ) if either a r The system is trllale if a ad. pprah : hek M()=[-I ] at =- M( ) a M( ) a - (M(-))= (M(-))= iff a ad Same lusi trllaility 7 geeral SI system (diagaliale) x x u The ave system is trllale if ad ly if the eigevalues are distit ad e f the i s is er 8 4

Example: x x u G [ ] α β β α ρ(g ) if ad detg ( ) ρ(g ) if either r The system is trllale if ad ( ) () 9 Therem: sider the pair m m Suppse that the eigevalues f i ad thse f j are disjit fr i j. The () is trllale iff ( i i ) is trllale fr all i. 5

6 Therem: Let () = p. The pair () is trllale iff G -p+: = [ -p ] has full rw rak. This is equivalet t G -p+g -p+ eig sigular ad t G -p+g -p+ > (psitive defiite.) Example: 3 =4 p=. ()== p. p 4 G The first 4 lums are LI. (G -p+)=4 = () trllale

Example: G p 3 3 3 The first 3 lums are LI. The 4 th is depedet the first 3. 3 Hee () is trllale. 3 =4 p=. ()=. 8 4 9 9 3 8 has full rw rak 4 9 3 Effet f equivalee trasfrmati Reall that equivalee trasfrmati a make the struture leaer ad simplify aalysis. Questi: Des similarity trasfrmati retai the trllaility prperty? Therem: The trllaility prperty is ivariat uder ay equivalee trasfrmati Prf: sider () with G =[. - ]. Let the trasfrmati matrix e P. The () (PP - P) G [ ] - [P PP P P - [P P P ] - P[ ] PG - P P] Sie P is sigular ρ( G ) ρ(g ) 4 7

Next Prlem: Oservaility 5 Oservaility: dual ept sider a -dimesial p-iput q-utput system: x x u; u System y x Du ssume that we kw the iput ad a measure the utput ut has aess t the state. Defiiti: The system is said t e servale if fr ay ukw iitial state x() there exists a fiite t > suh that x() a e exatly evaluated ver [t ] frm the iput u ad the utput y. Otherwise the system is said t e uservale. y 6 8

Duality etwee trllaility ad servaility Therem f duality: The pair () is trllale if ad ly if ( ) = ( ) is servale. x x u Dual systems ' y ' 7 Equivalet ditis fr servaility: ) The pair () is servale. ) W (t) is sigular fr sme t >. 3) The servaility matrix G has full lum rak i.e. (G ) =. 4) The matrix M ( λ) λi has full lum rak at every eigevalue f. 8 9

9 Therem: The pair () is servale if ad ly if q -q G has full lum rak where q=(). Therem: The seraility prperty is ivariat uder ay equivalee trasfrmati; Therem: sider the pair m m Suppse that the eigevalues f i ad thse f j are disjit fr.i j. The () is servale iff ( i i ) is servale fr all i.

S far we have leared trllaility Oservaility Next we will study aial dempsiti: t divide the state spae it trllale/utrllale servale/uservale suspaes aial Dempsiti sider a LTI system x x u y x Du Let = Px where P is sigular the where x u y Du PP - P P D D Reall that uder a equivalee trasfrmati all prperties suh as staility trllaility ad servaility are preserved. We als have G PG G G P Next we are gig t use equivalee trasfrmati t tai ertai speifi strutures whih reflet trllaility ad servaility.

trllaility dempsiti Reall G =[ - ]. Suppse that (G ) = <. The G has at mst LI lums. They frm a asis fr the rage spae f G. Therem: Suppse that (G ) = <. Let Q e a sigular matrix whse first lums are LI lums f G. Let P=Q -. The p PP P R R Mrever the pair ( ) is trlla le ad ( si ) D (si ) D See page 59 fr the prf. 3 Disussi: fter state trasfrmati the equivalet system is u The iput u has effet. This part f state is utrllale. The first su-system is trllale if =. If the (t ) e ( ) e t t e (t-τ) u( τ)d Give a desired value fr say d. If we let t (t-τ) v( t) e e d '(tt) ad u(t) 'e W (t )[e t W ( t) t The yu a verify that (t )= d. e t (t-τ) e τ v( t ) ( τ)d 'e d ] ' d 4

3 5 Example: u x x =3 p= -p+=. Oly eed t hek G G ale utrll 3 ) ( G Q P q] [ Q Let q is piked t make Q sigular P PQ Nte: the last lum f Q is differet frm the k (page 6). s a result Ā is differet frm that i the k whih is. 6 Therem: Suppse that (G ) = <. Let P e a sigular matrix whse first rws are LI rws f G. The p R R R P PP q D ) (si D ) ( ad servale is ) ( pair the Mrever si Oservaility dempsiti (fllws frm duality) G Reall Disussi: fter state trasfrmati the equivalet system is Du y u u may e affeted y ut has effet y r

Summary fr tday: trllaility Oservaility aial dempsiti trllale/utrllale Oservale/uservale Next Time: trllaility ad servaility tiued trllaility/servaility dempsiti Miimal realiati ditis fr Jrda frm ditis Parallel results fr disrete-time systems trllaility after samplig State feedak desig (itrduti) 7 Prlem Set #9. Is the fllwig state equati trllale? servale? x x u y x If t trllale redue it t a trllale e; If t servale redue it t a servale e.. Is the fllwig state equati trllale? servale? x x u y x If t trllale redue it t a trllale e; If t servale redue it t a servale e. 8 4