are positive, and the pair A, B is controllable. The uncertainty in is introduced to model control failures.

Similar documents
14. MRAC for MIMO Systems with Unstructured Uncertainties We consider affine-in-control MIMO systems in the form, x Ax B u f x t

9. MRAC Design for Affine-in-Control MIMO Systems

Some Different Perspectives on Linear Least Squares

= y and Normed Linear Spaces

7.0 Equality Contraints: Lagrange Multipliers

Relation (12.1) states that if two points belong to the convex subset Ω then all the points on the connecting line also belong to Ω.

( m is the length of columns of A ) spanned by the columns of A : . Select those columns of B that contain a pivot; say those are Bi

2/20/2013. Topics. Power Flow Part 1 Text: Power Transmission. Power Transmission. Power Transmission. Power Transmission

NONDIFFERENTIABLE MATHEMATICAL PROGRAMS. OPTIMALITY AND HIGHER-ORDER DUALITY RESULTS

Chapter 2: Descriptive Statistics

Department of Mathematics UNIVERSITY OF OSLO. FORMULAS FOR STK4040 (version 1, September 12th, 2011) A - Vectors and matrices

RECAPITULATION & CONDITIONAL PROBABILITY. Number of favourable events n E Total number of elementary events n S

SUBSEQUENCE CHARACTERIZAT ION OF UNIFORM STATISTICAL CONVERGENCE OF DOUBLE SEQUENCE

On EPr Bimatrices II. ON EP BIMATRICES A1 A Hence x. is said to be EP if it satisfies the condition ABx

VIII Dynamics of Systems of Particles

χ be any function of X and Y then

Numerical Analysis Formulae Booklet

such that for 1 From the definition of the k-fibonacci numbers, the firsts of them are presented in Table 1. Table 1: First k-fibonacci numbers F 1

A New Method for Solving Fuzzy Linear. Programming by Solving Linear Programming

Fairing of Parametric Quintic Splines

= lim. (x 1 x 2... x n ) 1 n. = log. x i. = M, n

XII. Addition of many identical spins

Fractional Integrals Involving Generalized Polynomials And Multivariable Function

for each of its columns. A quick calculation will verify that: thus m < dim(v). Then a basis of V with respect to which T has the form: A

Chain Rules for Entropy

Non-degenerate Perturbation Theory

A Convergence Analysis of Discontinuous Collocation Method for IAEs of Index 1 Using the Concept Strongly Equivalent

Robust Adaptive Asymptotic Tracking of Nonlinear Systems With Additive Disturbance

Mu Sequences/Series Solutions National Convention 2014

CS 2750 Machine Learning. Lecture 7. Linear regression. CS 2750 Machine Learning. Linear regression. is a linear combination of input components x

Assignment 5/MATH 247/Winter Due: Friday, February 19 in class (!) (answers will be posted right after class)

ANALYSIS ON THE NATURE OF THE BASIC EQUATIONS IN SYNERGETIC INTER-REPRESENTATION NETWORK

The Linear Probability Density Function of Continuous Random Variables in the Real Number Field and Its Existence Proof

Debabrata Dey and Atanu Lahiri

Exponential Generating Functions - J. T. Butler

ELEMENTS OF NUMBER THEORY. In the following we will use mainly integers and positive integers. - the set of integers - the set of positive integers

KURODA S METHOD FOR CONSTRUCTING CONSISTENT INPUT-OUTPUT DATA SETS. Peter J. Wilcoxen. Impact Research Centre, University of Melbourne.

CS 2750 Machine Learning Lecture 8. Linear regression. Supervised learning. a set of n examples

Coherent Potential Approximation

The Mathematical Appendix

{ }{ ( )} (, ) = ( ) ( ) ( ) Chapter 14 Exercises in Sampling Theory. Exercise 1 (Simple random sampling): Solution:

STK4011 and STK9011 Autumn 2016

Objectives. Learning Outcome. 7.1 Centre of Gravity (C.G.) 7. Statics. Determine the C.G of a lamina (Experimental method)

FIBONACCI-LIKE SEQUENCE ASSOCIATED WITH K-PELL, K-PELL-LUCAS AND MODIFIED K-PELL SEQUENCES

Born-Oppenheimer Approximation. Kaito Takahashi

Point Estimation: definition of estimators

PRACTICAL CONSIDERATIONS IN HUMAN-INDUCED VIBRATION

α1 α2 Simplex and Rectangle Elements Multi-index Notation of polynomials of degree Definition: The set P k will be the set of all functions:

Global Practical Output Tracking of Uncertain Nonlinear Systems By Smooth Output Feedback

Multiple Choice Test. Chapter Adequacy of Models for Regression

Sebastián Martín Ruiz. Applications of Smarandache Function, and Prime and Coprime Functions

University of Pavia, Pavia, Italy. North Andover MA 01845, USA

arxiv:math/ v1 [math.gm] 8 Dec 2005

A Family of Non-Self Maps Satisfying i -Contractive Condition and Having Unique Common Fixed Point in Metrically Convex Spaces *

Consumer theory. A. The preference ordering B. The feasible set C. The consumption decision. A. The preference ordering. Consumption bundle

C-1: Aerodynamics of Airfoils 1 C-2: Aerodynamics of Airfoils 2 C-3: Panel Methods C-4: Thin Airfoil Theory

Professor Wei Zhu. 1. Sampling from the Normal Population

Lecture Note to Rice Chapter 8

13. Parametric and Non-Parametric Uncertainties, Radial Basis Functions and Neural Network Approximations

MATH 247/Winter Notes on the adjoint and on normal operators.

Discrete Pseudo Almost Periodic Solutions for Some Difference Equations

Hájek-Rényi Type Inequalities and Strong Law of Large Numbers for NOD Sequences

The theoretical background of

d dt d d dt dt Also recall that by Taylor series, / 2 (enables use of sin instead of cos-see p.27 of A&F) dsin

Robust Stabilization of Uncertain Nonlinear Systems via Fuzzy Modeling and Numerical Optimization Programming

The Number of the Two Dimensional Run Length Constrained Arrays

A Variable Structure Model Reference Adaptive Control For MIMO Systems

THE PUBLISHING HOUSE PROCEEDINGS OF THE ROMANIAN ACADEMY, Series A, OF THE ROMANIAN ACADEMY Volume 9, Number 3/2008, pp

φ (x,y,z) in the direction of a is given by

Inequalities for Dual Orlicz Mixed Quermassintegrals.

Minimizing spherical aberrations Exploiting the existence of conjugate points in spherical lenses

1 Solution to Problem 6.40

A Conventional Approach for the Solution of the Fifth Order Boundary Value Problems Using Sixth Degree Spline Functions

VECTOR MECHANICS FOR ENGINEERS: Vector Mechanics for Engineers: Dynamics. In the current chapter, you will study the motion of systems of particles.

UNIT 2 SOLUTION OF ALGEBRAIC AND TRANSCENDENTAL EQUATIONS

Images of Linear Block Codes over Fq ufq vfq uvfq

Introduction to Matrices and Matrix Approach to Simple Linear Regression

9 U-STATISTICS. Eh =(m!) 1 Eh(X (1),..., X (m ) ) i.i.d

Discrete Mathematics and Probability Theory Fall 2016 Seshia and Walrand DIS 10b

This may involve sweep, revolution, deformation, expansion and forming joints with other curves.

Feature Selection: Part 2. 1 Greedy Algorithms (continued from the last lecture)

Unique Common Fixed Point of Sequences of Mappings in G-Metric Space M. Akram *, Nosheen

SUBCLASS OF HARMONIC UNIVALENT FUNCTIONS ASSOCIATED WITH SALAGEAN DERIVATIVE. Sayali S. Joshi

UNIT 7 RANK CORRELATION

Order Nonlinear Vector Differential Equations

Chapter 2 - Free Vibration of Multi-Degree-of-Freedom Systems - II

Complete Convergence and Some Maximal Inequalities for Weighted Sums of Random Variables

Ordinary Least Squares Regression. Simple Regression. Algebra and Assumptions.

Algorithms behind the Correlation Setting Window

An Expanded Method to Robustly Practically. Output Tracking Control. for Uncertain Nonlinear Systems

Chapter 8. Linear Momentum, Impulse, and Collisions

2006 Jamie Trahan, Autar Kaw, Kevin Martin University of South Florida United States of America

Chapter 9 Jordan Block Matrices

Random Variables. ECE 313 Probability with Engineering Applications Lecture 8 Professor Ravi K. Iyer University of Illinois

DATA DOMAIN DATA DOMAIN

Stability Analysis for Linear Time-Delay Systems. Described by Fractional Parameterized. Models Possessing Multiple Internal. Constant Discrete Delays

Partition and the Perfect Codes in the Additive Channel

Functions of Random Variables

1 Lyapunov Stability Theory

Strong Convergence of Weighted Averaged Approximants of Asymptotically Nonexpansive Mappings in Banach Spaces without Uniform Convexity

Transcription:

Lectue 4 8. MRAC Desg fo Affe--Cotol MIMO Systes I ths secto, we cosde MRAC desg fo a class of ult-ut-ult-outut (MIMO) olea systes, whose lat dyacs ae lealy aaetezed, the ucetates satsfy the so-called atchg codtos, ad f the full state s easuable, (.e., avalable o-le as the syste outut). Secfcally, we cosde th ode MIMO systes the fo, ABu f (8.1) whee R s the syste state, u R s the cotol ut, B R s kow, whle A R ad R ae ukow atces. I addto, t s assued that s dagoal, ts eleets ae ostve, ad the a A, B s cotollable. he ucetaty s toduced to odel cotol falues. he ukow, ossbly olea, fucto : f R R eesets the so-called syste atched ucetaty. It s assued that the fucto ca be wtte as a lea cobato of N kow bass fuctos, wth ukow costat coeffcets. f (8.) N I (8.), R N s the ukow costat at, ad R deotes the kow egesso vecto. I ode to guaatee estece ad uqueess of the syste tajectoes, t s assued that s locally Lschtz. he cotol objectve of the MIMO tackg oble s to choose the cotol ut u such that all sgals the closed-loo syste ae bouded, ad the state follows ef R the state of a efeece odel, secfed by the LI syste, ef Aef ef Bef t (8.3) whee R s Huwtz, R, ad t R s a bouded efeece ut Aef Bef (eteal coad). Note that the efeece odel dyacs ad the eteal ut t ust be chose so that ef t eesets the desed tajectoy fo t to follow. I suay, gve a bouded coad t, the cotol ut u eeds to be chose such that the tackg eo globally asytotcally teds to zeo. l t t 0 (8.4) t If the atces A ad wee kow, oe could have aled the cotol law, u (8.5) ad obta the closed-loo syste: AB B (8.6) ef 3

Coag (8.6) wth the desed dyacs (8.3), t follows that the deal (ukow) at gas ust be chose to satsfy the so-called atchg codtos: AB Aef (8.7) B Bef Assug that the atchg codtos take lace, t s easy to see that the closed-loo syste s the sae as the efeece odel, ad cosequetly, asytotc (eoetal) tackg s acheved fo ay bouded efeece ut sgal t. Reak 8.1 Gve the atces A, B,, Aef, Bef, o, ay est to satsfy the atchg codtos (8.7) dcatg that the cotol law (8.5) ay ot have eough stuctual fleblty to eet the cotol objectve. Ofte actce, the stuctue of A s kow, ad the efeece odel atces Aef, B ef ae chose so that (8.7) has a soluto fo,. Assug that, (8.7) est, cosde the followg cotol law: u (8.8) whee,, N R R R ae the estates of the deal ukow atces,,, esectvely. he estated atces wll be geeated o-le. Substtutg (8.8) to (8.1), the closed-loo syste dyacs ca be wtte. AB B (8.9) Subtactg (8.3) fo (8.9), closed-loo dyacs of the desoal tackg eo et t t ca be obtaed. vecto ef e AB B A B (8.10) ef ef ef Usg atchg codtos (8.7) futhe yelds: e A ef B Aef ef B B (8.11) A ef eb Let,, ad eeset the aaete estato eos. I tes of the latte, the tackg eo dyacs becoes: e Aef eb (8.1) Vecto ad at os Befoe oceedg ay futhe, ecall that gve a at Fobeus o s defed by A a R j, the at 33

wth j A t A A a (8.13) F t deotg the tace oeato. O the othe had, gve ay vecto -o, the duced at o s defed by A, j A su (8.14) 0 Collecto of Facts about vecto ad at os, (ove t). Fo vecto 1-o 1, the duced at o s equal to the au 1 absolute colu su, that s: A a 1 aj. 1 Fo vecto -o 1 j 1 sgula value of A, that s: A A Fo vecto -o absolute ow su, that s: 1, the duced at o s equal to the au a A. a, the duced at o s equal to the au a a. 1 j1 he duced at o satsfes: A A, ad fo ay two coatbly desoed atces, A ad B, oe also has: A B A B. he Fobeus o s ot a duced o of ay vecto o, but t s coatble wth the -o the sese that: A A. F Fo ay two coatbly desoed atces A ad B, the Fobeus e oduct s defed as: AB, tace F A B. Accodg to the Schwatz equalty oe has: tace A B A, B A B j F F F Fo ay two co desoal vectos a ad b, the tace detty takes lace: a b ba t Let 0, 0, 0. Gog back to aalyzg the tackg eo dyacs (8.1), cosde the Lyauov fucto caddate: 1 1 1 V e,,, e Pet (8.15) whee P P 0 satsfes the algebac Lyauov equato, PA A P Q (8.16) ef ef 34

fo soe QQ 0. he the te devatve of V, evaluated alog the tajectoes of (8.1), ca be calculated. 1 1 1 V e Pe e Pe t ef ef A eb Pe ep A eb 1 1 1 t e A PPA ee PB ef ef 1 1 1 t Usg (8.16), futhe yelds: 1 V e Qe e PB t 1 epb t 1 epb t Usg the tace detty, oe gets epb t epb b b a a epb t epb b b a a epb t epb a b b a Substtutg (8.19) to (8.18), esults : 1 V e Qe t e PB 1 1 t e PB t e PB If the adatve laws ae chose as, e PB t e PB epb the the te-devatve of V becoes egatve se-defte. (8.17) (8.18) (8.19) (8.0) (8.1) 35

V e Qe0 (8.) heefoe, the closed-loo eo dyacs ae stable, that s the tackg eo et ad the aaete estato eos t, t, t ae ufoly bouded fuctos of te. Cosequetly, the aaete estates t, t, t ae also bouded. Sce t s bouded the ef t ad ef t ae bouded. Hece, the syste state t s bouded ad the cotol ut ut (8.8) s bouded. he latte les that t s bouded ad thus et s bouded. Futheoe, the d te devatve of V t V e Qee Qe (8.3) s bouded, ad so V t s a ufoly cotuous fucto of te. he latte couled wth the facts that V t s lowe bouded ad V t 0 les (Babalat s Lea) that. hus, et lv t 0 t l 0 ad the MIMO tackg oble s solved. t Reak 8. (Hoewok: Pove ths stateet) If soe of the dagoal eleets of the ukow dagoal at ae egatve ad the sgs of all of the ae kow, the the adatve laws e PBsg t e PBsg (8.4) e PBsg solve the MIMO tackg oble, whee sg dagsg 1,, sg. 36