Lecture 1: Contents of the course. Advanced Digital Control. IT tools CCSDEMO

Similar documents
Laplace Transforms. Examples. Is this equation differential? y 2 2y + 1 = 0, y 2 2y + 1 = 0, (y ) 2 2y + 1 = cos x,

Lecture 1 Overview. course mechanics. outline & topics. what is a linear dynamical system? why study linear systems? some examples

Recursive Least-Squares Fixed-Interval Smoother Using Covariance Information based on Innovation Approach in Linear Continuous Stochastic Systems

LAPLACE TRANSFORM AND TRANSFER FUNCTION

5. Response of Linear Time-Invariant Systems to Random Inputs

Chapter 1 Fundamental Concepts

ADDITIONAL PROBLEMS (a) Find the Fourier transform of the half-cosine pulse shown in Fig. 2.40(a). Additional Problems 91

( ) ( ) if t = t. It must satisfy the identity. So, bulkiness of the unit impulse (hyper)function is equal to 1. The defining characteristic is

10. State Space Methods

Sampling of Linear Systems

6.003 Homework #9 Solutions

EE 315 Notes. Gürdal Arslan CLASS 1. (Sections ) What is a signal?

Chapter 3 Boundary Value Problem

Simulation-Solving Dynamic Models ABE 5646 Week 2, Spring 2010

6.003 Homework #9 Solutions

KEY. Math 334 Midterm III Winter 2008 section 002 Instructor: Scott Glasgow

DEPARTMENT OF ELECTRICAL AND ELECTRONIC ENGINEERING EXAMINATIONS 2008

MAE143A Signals & Systems - Homework 2, Winter 2014 due by the end of class Thursday January 23, 2014.

ψ ( t) = c n ( t) t n ( )ψ( ) t ku t,t 0 ψ I V kn

Differential Equations

STATE-SPACE MODELLING. A mass balance across the tank gives:

Solutions of Sample Problems for Third In-Class Exam Math 246, Spring 2011, Professor David Levermore

Hamilton- J acobi Equation: Explicit Formulas In this lecture we try to apply the method of characteristics to the Hamilton-Jacobi equation: u t

Lecture 20: Riccati Equations and Least Squares Feedback Control

An recursive analytical technique to estimate time dependent physical parameters in the presence of noise processes

Homework 6 AERE331 Spring 2019 Due 4/24(W) Name Sec. 1 / 2

Continuous Time. Time-Domain System Analysis. Impulse Response. Impulse Response. Impulse Response. Impulse Response. ( t) + b 0.

Spring Ammar Abu-Hudrouss Islamic University Gaza

Augmented Reality II - Kalman Filters - Gudrun Klinker May 25, 2004

Zürich. ETH Master Course: L Autonomous Mobile Robots Localization II

KEY. Math 334 Midterm III Fall 2008 sections 001 and 003 Instructor: Scott Glasgow

Some Basic Information about M-S-D Systems

Ordinary differential equations. Phys 750 Lecture 7

Signals and Systems Profs. Byron Yu and Pulkit Grover Fall Midterm 1 Solutions

h[n] is the impulse response of the discrete-time system:

6.2 Transforms of Derivatives and Integrals.

Using the Kalman filter Extended Kalman filter

Application of a Stochastic-Fuzzy Approach to Modeling Optimal Discrete Time Dynamical Systems by Using Large Scale Data Processing

ψ ( t) = c n ( t ) n

B Signals and Systems I Solutions to Midterm Test 2. xt ()

Notes on Kalman Filtering

+ γ3 A = I + A 1 (1 + γ + γ2. = I + A 1 ( t H O M E W O R K # 4 Sebastian A. Nugroho October 5, 2017

L07. KALMAN FILTERING FOR NON-LINEAR SYSTEMS. NA568 Mobile Robotics: Methods & Algorithms

Intermediate Differential Equations Review and Basic Ideas

Sensors, Signals and Noise

Modal identification of structures from roving input data by means of maximum likelihood estimation of the state space model

Vanishing Viscosity Method. There are another instructive and perhaps more natural discontinuous solutions of the conservation law

Probabilistic Robotics

RC, RL and RLC circuits

4/9/2012. Signals and Systems KX5BQY EE235. Today s menu. System properties

Section 3.5 Nonhomogeneous Equations; Method of Undetermined Coefficients

Q1) [20 points] answer for the following questions (ON THIS SHEET):

DISCRETE GRONWALL LEMMA AND APPLICATIONS

Introduction to Mobile Robotics

Math Final Exam Solutions

CONTRIBUTION TO IMPULSIVE EQUATIONS

Lab 10: RC, RL, and RLC Circuits

EE363 homework 1 solutions

Signal and System (Chapter 3. Continuous-Time Systems)

Announcements. Recap: Filtering. Recap: Reasoning Over Time. Example: State Representations for Robot Localization. Particle Filtering

Algorithmic Trading: Optimal Control PIMS Summer School

Georey E. Hinton. University oftoronto. Technical Report CRG-TR February 22, Abstract

Institute for Mathematical Methods in Economics. University of Technology Vienna. Singapore, May Manfred Deistler

EE 435 Lecture 42. Phased Locked Loops and VCOs

( ) = Q 0. ( ) R = R dq. ( t) = I t

Determination of the Sampling Period Required for a Fast Dynamic Response of DC-Motors

Problemas das Aulas Práticas

Chapter 2. First Order Scalar Equations

6.003: Signals and Systems

Class Meeting # 10: Introduction to the Wave Equation

2.160 System Identification, Estimation, and Learning. Lecture Notes No. 8. March 6, 2006

Homework-8(1) P8.3-1, 3, 8, 10, 17, 21, 24, 28,29 P8.4-1, 2, 5

Two Popular Bayesian Estimators: Particle and Kalman Filters. McGill COMP 765 Sept 14 th, 2017

Vehicle Arrival Models : Headway

RL Lecture 7: Eligibility Traces. R. S. Sutton and A. G. Barto: Reinforcement Learning: An Introduction 1

EE 224 Signals and Systems I Complex numbers sinusodal signals Complex exponentials e jωt phasor addition

Kinematics and kinematic functions

Voltage/current relationship Stored Energy. RL / RC circuits Steady State / Transient response Natural / Step response

From Particles to Rigid Bodies

Representing a Signal. Continuous-Time Fourier Methods. Linearity and Superposition. Real and Complex Sinusoids. Jean Baptiste Joseph Fourier

dt = C exp (3 ln t 4 ). t 4 W = C exp ( ln(4 t) 3) = C(4 t) 3.

Lectures 29 and 30 BIQUADRATICS AND STATE SPACE OP AMP REALIZATIONS. I. Introduction

2.7. Some common engineering functions. Introduction. Prerequisites. Learning Outcomes

Outline Chapter 2: Signals and Systems

Utility maximization in incomplete markets

CHAPTER 2 Signals And Spectra

Notes 04 largely plagiarized by %khc

Ch.1. Group Work Units. Continuum Mechanics Course (MMC) - ETSECCPB - UPC

CHBE320 LECTURE IV MATHEMATICAL MODELING OF CHEMICAL PROCESS. Professor Dae Ryook Yang

Block Diagram of a DCS in 411

Wall. x(t) f(t) x(t = 0) = x 0, t=0. which describes the motion of the mass in absence of any external forcing.

Module 4: Time Response of discrete time systems Lecture Note 2

(b) (a) (d) (c) (e) Figure 10-N1. (f) Solution:

Lecture 13 RC/RL Circuits, Time Dependent Op Amp Circuits

HOMEWORK # 2: MATH 211, SPRING Note: This is the last solution set where I will describe the MATLAB I used to make my pictures.

Chapter 7 Response of First-order RL and RC Circuits

Deep Learning: Theory, Techniques & Applications - Recurrent Neural Networks -

MA 366 Review - Test # 1

Applications in Industry (Extended) Kalman Filter. Week Date Lecture Title

Physics 235 Chapter 2. Chapter 2 Newtonian Mechanics Single Particle

Transcription:

Goals of he course Lecure : Advanced Digial Conrol To beer undersand discree-ime sysems To beer undersand compuer-conrolled sysems u k u( ) u( ) Hold u k D-A Process Compuer y( ) A-D y ( ) Sampler y k y k Conens of he course Sampling of sysems and signals Analysis of discree-ime sysems I/O models More abou z-ransform and δ -ransform Hybrid sysems Design mehods Reference values and inegraors I/O models and connecion o sae-space models Opimal conrol, minimum variance and LQG IT ools Course informaion, available ools CCSDEMO Malab and Simulink ool for he course WWW a hp://www.conrol.lh.se/educaion/ DocoraeProgram/advanced-digial-conrol/ Figures generaed wih macros available Malab and Simulink /home/ccs/malab/ccs5/chaperx Ineracive ool ccsdemo

Sampling of CCS Example s+b s 2 +a s+a 2 for differen sampling inervals Sampling of signals and sysems Inroducion o compuer-conrolled sysems Sampling of signals Sampling of sysems Sae-space models -oupu models Discree-ime sysems Disregard inersample behavior Problems wih sampled-daa sysems dependence u A-D Compuer Clock D-A y s (b) A naive approach Conrol of he arm of a disk drive Coninuous ime conroller Discree ime conroller G(s)= k Js 2 U(s)= bk a U c(s) K s+b s+a Y(s) u( k )=K( b a u c( k ) y( k )+x( k )) x( k +h)=x( k )+h((a b)y( k ) ax( k ))

Conrol of he double inegraor Clock Algorihm Conrol of he double inegraor Shor sampling period Sampling periodh=.2/ω 5.5 y: = adin(in2) {read process value} u:=k*(a/b*uc-y+x) dou(u) {oupu conrol signal} newx:=x+h*((b-a)*y-b*x).5 5 (ω ) Conrol of he double inegraor Increased sampling period a)h=.5/ω b)h=.8/ω 5.5 (b) 5.5 Dead-bea conrolh=.4/ω Beer performance? u( k )= u c ( k )+ u c ( k ) s y( k ) s y( k ) r u( k ) Posiion Velociy 5.5.5 5.5 5 5 (ω ).5 5 5 (ω ).5 5 (ω )

Sinusoidal measuremen noise Sinusoidal measuremen noise con d Coninuous-ime Discree-ime Enlarge he scale.2.2 2.2.2 2.2.2 2 Measured oupu (b).2.2 2.2.2 2.2.2 2 Sampling creaes new frequencies! ω sampled = ω±nω s Measured oupu Measured oupu.2.2.2.2 5 5 The sampling period is oo long compared wih he noise Imporan o filer before sampling! Problems Higher-order harmonics u y s A-D Compuer D-A Sampling of sysems Look a he sysem from he poin of view of he compuer Clock Clock { } {u( k )} u() y() y( k) D-A Sysem A-D (b) (c) Sampl. oupu Con. oupu 5 5 5 Zero-order-hold sampling of a sysem Compuaional issues Soluion of he sysem equaion Inverse of sampling Sampling of a sysem wih ime delay Inersample behavior

The idea: Sampling a coninuous-ime sysem Le he inpu be piecewise consan Look a he sampling poins only Use lineariy and calculae sep responses Sysem descripion dx d =Ax()+Bu() y()=cx()+du() Sampling a coninuous-ime sysem, con d Solve he sysem equaion (Calculae he sep response, wih iniial value) x()=e A( k) x( k )+ k e A( s ) Bu(s )ds =e A( k) x( k )+ e A( s ) ds Bu( k ) k k =e A( k) x( k )+ e As dsbu( k ) = Φ(, k )x( k )+Γ(, k )u( k ) The general case Periodic sampling x( k+ )= Φ( k+, k )x( k )+ Γ( k+, k )u( k ) y( k )=Cx( k )+Du( k ) k+ k Γ( k+, k )= e As dsb Φ( k+, k )=e A( k+ k ) Assume periodic sampling, i.e. k =k h, hen x(kh+h)=φx(kh)+γu(kh) y(kh)=cx(kh)+du(kh) NOTE: -invarian linear sysem! Φ=e Ah Γ= h e As dsb I hus follows ha Properies of Φ and Γ Φ=e Ah Γ= h e As dsb dφ() =AΦ()= Φ()A d dγ() = Φ()B d Φ and Γ saisfy d Φ() Γ() Φ() Γ() = d I I A B Φ(h) and Γ(h) can be obained from he block marix } Φ(h) Γ(h) =exp{ A B h I

How o compue Φ and Γ? Numerical calculaion in Malab Series expansion of he marix exponenial. The Laplace ransform ofexp(a) is(si A). Cayley-Hamilon s heorem. Symbolic compuer algebra Inverse of sampling Problem: Is i always possible o find a coninuous-ime sysem ha corresponds o a discree-ime sysem? Exisence (Example 2.4 Firs order sysem) Non-uniqueness (Example 2.5 Harmonic oscillaor) One way is Ψ= h e As ds=ih+ Ah2 2! The marices Φ and Γ are given by + A2 h 3 3! + Φ=I+AΨ=I+Ah+ (Ah)2 2! + (Ah)3 3! + Γ=ΨB Ierae he equaions Soluion of he sysem equaion x(k+)=φx(k)+γu(k) y(k)=cx(k)+du(k) x(k)=φ k k x(k )+ Φ k k Γu(k ) + + Γu(k ) = Φ k k x(k )+ k j=k Φ k j Γu(j) k y(k)=cφ k k x(k )+ CΦ k j Γu(j)+Du(k) j=k k =CΦ k k x(k )+ h(k j)u(j) j=k Iniial value + Influence of inpu signal Sampling of sysem wih ime delay u( ) Delayed signal τ kh h kh kh + h kh + 2h dx() =Ax()+Bu( τ) d x(kh+h)=e Ah x(kh) + kh+h kh e A(kh+h s ) Bu(s τ)ds

delay con d Spli over piece-wise consan pars kh+h kh kh+τ = where kh e A(kh+h s ) Bu(s τ)ds kh+h e A(kh+h s ) Bds u(kh h)+ e A(kh+h s ) Bds u(kh) kh+τ = Γ u(kh h)+γ u(kh) x(kh+h)=φx(kh)+γ u(kh)+ Γ u(kh h) Φ=e Ah Γ = h τ τ e As dsb Γ =e A(h τ) e As dsb Sae-space model x(kh+h) = Φ Γ x(kh) u(kh) u(kh h) + Γ u(kh) I Noice:rexra sae variablesu(kh h) Longer ime delays τ=(d )h+ τ < τ h x(kh+h)=φx(kh)+γ u(kh (d )h) + Γ u(kh dh) Example Double inegraor wih delay Consider he double inegraor wih delay τ Φ=e Ah = h τ Γ =e A(h τ) e As dsb= h τ (h = τ τ ) 2 τ h τ (h τ) 2 Γ = e As dsb= 2 h τ τ 2 2 τ x(kh+h)=φx(kh)+γ u(kh h)+ Γ u(kh) Inner ime-delay u y u 2 y S e sτ S 2 Le he sysem be described by he equaions S : dx () =A x ()+B u() d y ()=C x ()+D u() S 2 : dx 2() =A 2 x 2 ()+B 2 u 2 () d u 2 ()=y ( τ) u() piecewise consan over he sampling inerval h Find a sae-space descripion of he sampled sysem

Inner ime-delay con d Sampling wih τ= and sampling periodh x (kh+h) = Φ (h) x (kh) + Γ (h) u(kh) x 2 (kh+h) Φ 2 (h) Φ 2 (h) x 2 (kh) Γ 2 (h) where Φ i ()=e A i Γ ()= e A s B ds Wienmark (985), generalized in Bernhardsson (993) where Φ i ()=e A i Φ 2 ()= Γ ()= Inner ime-delay Theorem x (kh+h)= Φ (h)x (kh)+γ (h)u(kh) x 2 (kh+h)= Φ 2x (kh h)+φ 2 (h)x 2 (kh) + Γ 2u(kh h)+γ 2 (h τ)u(kh) i=,2 e A 2s B 2 C e A ( s) ds e A s B ds Γ 2 ()= Φ 2 = Φ 2(h)Φ (h τ) e A 2s B 2 C Γ ( s)ds Γ 2 = Φ 2(h)Γ (h τ)+φ 2 (h τ)γ (τ)+φ 2 (h τ)γ 2 (τ) Sample he delay-free sysem wihh,h τ, and τ! -oupu models Solving he sysem equaion gives y(k)=cφ k k x(k )+ k Higher order difference equaion Pulse-response (weighing) funcion h(k)= CΦ k Γ j=k CΦ k j Γu(j)+Du(k) k< D k= k Summary New phenomena Sysem wih ime delay Soluion of sysem equaion -oupu models y(k)=cφ k k x(k )+ k j=k h(k j)u(j)