Mathematics for Control Theory

Similar documents
Nonlinear Control. Nonlinear Control Lecture # 6 Passivity and Input-Output Stability

Lecture 8. Chapter 5: Input-Output Stability Chapter 6: Passivity Chapter 14: Passivity-Based Control. Eugenio Schuster.

Passivity-based Control of Euler-Lagrange Systems

Stability of Parameter Adaptation Algorithms. Big picture

Introduction to Nonlinear Control Lecture # 4 Passivity

Dissipativity. Outline. Motivation. Dissipative Systems. M. Sami Fadali EBME Dept., UNR

EG4321/EG7040. Nonlinear Control. Dr. Matt Turner

Balancing of Lossless and Passive Systems

IMPLICATIONS OF DISSIPATIVITY AND PASSIVITY IN THE DISCRETE-TIME SETTING. E.M. Navarro-López D. Cortés E. Fossas-Colet

On the passivity of general nonlinear systems

ESC794: Special Topics: Model Predictive Control

Nonlinear Control Lecture 7: Passivity

Dissipative Systems Analysis and Control

Lecture 5 Input output stability

Semidefinite Programming Duality and Linear Time-invariant Systems

Nonlinear Control. Nonlinear Control Lecture # 25 State Feedback Stabilization

Output Feedback and State Feedback. EL2620 Nonlinear Control. Nonlinear Observers. Nonlinear Controllers. ẋ = f(x,u), y = h(x)

Solution of Linear State-space Systems

Passivity Indices for Symmetrically Interconnected Distributed Systems

EE363 homework 8 solutions

Engineering Tripos Part IIB Nonlinear Systems and Control. Handout 4: Circle and Popov Criteria

arxiv: v3 [math.oc] 1 Sep 2018

7.1 Linear Systems Stability Consider the Continuous-Time (CT) Linear Time-Invariant (LTI) system

A new passivity property of linear RLC circuits with application to Power Shaping Stabilization

Equilibrium-Independent Passivity: a New Definition and Implications

MCE/EEC 647/747: Robot Dynamics and Control. Lecture 8: Basic Lyapunov Stability Theory

Tracking and Disturbance Rejection for Passive Nonlinear Systems

Control Systems I. Lecture 2: Modeling. Suggested Readings: Åström & Murray Ch. 2-3, Guzzella Ch Emilio Frazzoli

Control Systems I. Lecture 2: Modeling and Linearization. Suggested Readings: Åström & Murray Ch Jacopo Tani

DISSIPATION CONTROL OF AN N-SPECIES FOOD CHAIN SYSTEM

Nonlinear Control Lecture # 14 Input-Output Stability. Nonlinear Control

MCE693/793: Analysis and Control of Nonlinear Systems

Converse Lyapunov theorem and Input-to-State Stability

Towards Power-Based Control Strategies for a Class of Nonlinear Mechanical Systems Rinaldis, Alessandro de; Scherpen, Jacquelien M.A.

Exam. 135 minutes, 15 minutes reading time

A differential Lyapunov framework for contraction analysis

u e G x = y linear convolution operator. In the time domain, the equation (2) becomes y(t) = (Ge)(t) = (G e)(t) = Z t G(t )e()d; and in either domains

Control of Robotic Manipulators with Input/Output Delays

Outline. Input to state Stability. Nonlinear Realization. Recall: _ Space. _ Space: Space of all piecewise continuous functions

We are devoted to advance in the study of the behaviour of nonlinear discrete-time systems by means of its energy properties.

EEE582 Homework Problems

Interconnection and Damping Assignment Approach for Reliable PM Synchronous Motor Control

Georgia Institute of Technology Nonlinear Controls Theory Primer ME 6402

On the Design of Nonlinear Controllers for Euler-Lagrange Systems

Copyrighted Material. 1.1 Large-Scale Interconnected Dynamical Systems

Modeling and Analysis of Dynamic Systems

Lyapunov Stability Theory

Dynamical Systems & Lyapunov Stability

QUANTITATIVE L P STABILITY ANALYSIS OF A CLASS OF LINEAR TIME-VARYING FEEDBACK SYSTEMS

Robust Stabilizing Output Feedback Nonlinear Model Predictive Control by Using Passivity and Dissipativity

Passivity-based Adaptive Inventory Control

Analysis and Control of Multi-Robot Systems. Elements of Port-Hamiltonian Modeling

Lyapunov-based methods in control

Applied Nonlinear Control

Stabilization and Passivity-Based Control

Chapter One. Introduction

TTK4150 Nonlinear Control Systems Solution 6 Part 2

QSR-Dissipativity and Passivity Analysis of Event-Triggered Networked Control Cyber-Physical Systems

Exam in Systems Engineering/Process Control

Linear Control Systems Lecture #3 - Frequency Domain Analysis. Guillaume Drion Academic year

Nonlinear Control. Nonlinear Control Lecture # 8 Time Varying and Perturbed Systems

Linear Matrix Inequalities in Robust Control. Venkataramanan (Ragu) Balakrishnan School of ECE, Purdue University MTNS 2002

ROBUST PASSIVE OBSERVER-BASED CONTROL FOR A CLASS OF SINGULAR SYSTEMS

I. D. Landau, A. Karimi: A Course on Adaptive Control Adaptive Control. Part 9: Adaptive Control with Multiple Models and Switching

Dissipative Systems Analysis and Control, Theory and Applications: Addendum/Erratum

Modeling & Control of Hybrid Systems Chapter 4 Stability

16.30/31, Fall 2010 Recitation # 13

STABILITY AND STABILIZATION OF A CLASS OF NONLINEAR SYSTEMS WITH SATURATING ACTUATORS. Eugênio B. Castelan,1 Sophie Tarbouriech Isabelle Queinnec

CDS Solutions to the Midterm Exam

Subject: Optimal Control Assignment-1 (Related to Lecture notes 1-10)

Marcus Pantoja da Silva 1 and Celso Pascoli Bottura 2. Abstract: Nonlinear systems with time-varying uncertainties

Raktim Bhattacharya. . AERO 422: Active Controls for Aerospace Vehicles. Dynamic Response

FRTN 15 Predictive Control

D(s) G(s) A control system design definition

Exam. 135 minutes + 15 minutes reading time

When Gradient Systems and Hamiltonian Systems Meet

Quadratic Stability of Dynamical Systems. Raktim Bhattacharya Aerospace Engineering, Texas A&M University

Feedback stabilisation with positive control of dissipative compartmental systems

Minimum-Phase Property of Nonlinear Systems in Terms of a Dissipation Inequality

MCE693/793: Analysis and Control of Nonlinear Systems

On the PDEs arising in IDA-PBC

5. Observer-based Controller Design

CDS Solutions to Final Exam

KINETIC ENERGY SHAPING IN THE INVERTED PENDULUM

Nonlinear Control Systems

Passivity-based Stabilization of Non-Compact Sets

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

Introduction to Nonlinear Control Lecture # 3 Time-Varying and Perturbed Systems

State Regulator. Advanced Control. design of controllers using pole placement and LQ design rules

Time-Invariant Linear Quadratic Regulators Robert Stengel Optimal Control and Estimation MAE 546 Princeton University, 2015

Can Thermodynamics Be Used to Design Control Systems?

Nonlinear Control Systems

Model reduction for linear systems by balancing

Chap. 3. Controlled Systems, Controllability

The Important State Coordinates of a Nonlinear System

arxiv: v1 [cs.sy] 20 Nov 2017

Control of the Inertia Wheel Pendulum by Bounded Torques

Automatic Control 2. Nonlinear systems. Prof. Alberto Bemporad. University of Trento. Academic year

Time-Invariant Linear Quadratic Regulators!

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

Transcription:

Mathematics for Control Theory Outline of Dissipativity and Passivity Hanz Richter Mechanical Engineering Department Cleveland State University

Reading materials Only as a reference: Charles A. Desoer and Mathukumalli Vidyasagar (2009) [1975], Feedback Systems: Input-Output Properties, Classics in Applied Mathematics, SIAM, ISBN-13: 978-0898716702 (chapter on passivity) Romeo Ortega, Antonio Loria, Per Johan Nicklasson and Hebertt Sira-Ramirez, (1998), Passivity-Based Control of Euler-Lagrange Systems, Springer, ISBN 978-1-4471-3603-3. 2 / 17

Energy conservation u ẋ = Ax+Bu y = Cx y Define u = F, y = ẋ (we know power=uy). Then in a time interval [t 1,t 2 ] we expect: energy stored in m, k = energy delivered by F - energy dissipated in b 3 / 17

Dissipation Inequality: Physical Systems Since the dissipated energy is nonnegative, we can also say: Energy delivered energy stored In this system, E = 1 2 mẋ2 + 1 2 kx2 and t2 Since energy is nonnegative: t 1 u(τ)y(τ)dτ E(t 2 ) E(t 1 ) t2 t 1 u(τ)y(τ)dτ E(t 1 ) = β Passivity is defined using just a constant β on the right-hand side (no need to involve a storage function). 4 / 17

Dissipation Inequality: Other Systems Given an arbitrary LTI system u ẋ = Ax+Bu y = Cx y Is there an energy-like function E(x) so that t2 for any [t 1,t 2 ] and any input u(t)? t 1 u(τ)y(τ)dτ E(x(t 2 )) E(x(t 1 )) 5 / 17

Notation Input and output signals are assumed to take values in R m and R p for some m,p 1. These signals are assumed to be in the L 2e or L 2 classes, as appropriate. Notations like L m 2e could be used, but the superscript will be omitted when the dimension of the value space is clear from the context. 6 / 17

Definition of Passivity System: Σ : { ẋ = f(x,u), x(0) = x0 R n y = h(x,u) with u and y are L 2e signals taking values in R m and R m, respectively. System Σ is passive if there exists a storage function H : R n R 0+ s.t. u y T T 0 u T (t)y(t)dt H(x(T)) H(x(0)) for all u L 2e, all T 0 and all x 0 R n. 7 / 17

LTI Systems: Kalman-Yakubovich-Popov Lemma System: Σ : { ẋ = Ax+Bu, x(0) = x0 R n y = Cx The system is passive with storage function V(x) = 1 2 xt Px iff P = P T 0 and Q = Q T 0 s.t. A T P +PA = Q C = B T P This result (KY Lemma) was a breakthrough in systems theory in the 1960 s. It was cast in the slidework of controls by Kalman. Yakubovich and Popov had worked on the problem previously. 8 / 17

Properties of Passive LTI Systems Passivity in LTI systems implies stability. Also, passivity implies and is implied by the positive-real property: with no poles of G(s) in Re(s) 0 Geometrically: G(jw)+G( jw) 0 w R The Nyquist plot of G(s) is contained in the right-half plane, that is, the phase of G(s) lies within [ 90, 90 ]. Also, the relative degree of G must be 1. These properties are fundamental to establish stability robustness (infinite gain margin, good phase margin) and stability with nonlinear feedback interconnections. 9 / 17

Example: Using the KY Lemmma If a LTI system is defined by A,B and it s up to us to select outputs for feedback, (C matrix), we can make the system passive through the KY lemma. Example: lindemo.mdl, lin_pass_demo.m Set up random but stable 3rd order system A,B. Compute C to render the system passive via KY Check PR property with a Nyquist plot. Verify dissipation inequality Check robustness against any positive gain uncertainties, some phase shift via delay. 10 / 17

Passivity in Nonlinear Systems Besides linear systems born from the KY lemma, which others are passive? There are nonlinear versions of the KYP lemma (Moylan, TAC 74; Savkin, IFAC 93). Euler-Lagrange systems viewed as control systems are passive: d dt L q L q + F q = Mu+Qζ q R n is the vector of generalized coordinates L is the Lagrangian F is a dissipation function (think b q 2 /2) u is the control, ζ are disturbances and M and Q are constant matrices. In mechanical systems, the Lagrangian can be chosen as L(q, q) T (q, q) V(q) (difference between kinetic and potential energies). 11 / 17

Passivity Implies Stability in Nonlinear Systems Definitions: 1. Output strict passivity (OSP): u y T δ o y 2 T +β o for some δ o > 0,β o 2. Zero-state detectability (ZSD): A state-space system is ZSD if the following implication is true: y(t) = 0 t 0 = lim t x(t) = 0 12 / 17

Passivity Implies Lyapunov Stability... Proposition (Hill and Moylan, 1976; van der Schaft, 1996): Consider the system: Σ : { ẋ = f(x)+g(x,u), x(0) = x0 R n y = h(x,u) with u and y taking values in R m. If Σ is OSP with storage function H > 0 and it is ZSD, then x = 0 is a local asymptotically stable equilibrium point. Further, if H is radially unbounded, the stability is global. H is radially unbounded if lim x H(x) = 13 / 17

Passivity theorem A feedback interconnection of passive systems maintains stability under some conditions. u 1 + e 1 H 1 H 2 e 2 + + u 2 1. Input strict passivity (ISP): u y T δ i u 2 T +β i for some δ i > 0,β i This feedback interconnection can be used to capture many control systems. Additional definitions: 2. Finite gain (we studied something equivalent with the SGT). y T γ u T +β for some γ > 0,β Properties must hold for all inputs u L 2e and all T 0. 14 / 17

A version of the passivity theorem (Desoer and Vidyasagar) Suppose H 1 and H 2 map L 2e into itself. Assume that for any u 1 and u 2 in L 2, there are solutions e 1 and e 2 in L 2e. Suppose that: 1. H 2 satisfies a passivity-like inequality relative to the output: H 2 e 2 e 2 T δ o H 2 e 2 2 T +β o for some δ o,β o 2. H 1 satisfies a passivity-like inequality relative to the input: e 1 H 1 e 1 T δ i e 1 2 T +β i for some δ i,β i 3. H 1 has finite gain. Then the feedback system is internally L 2 -stable if δ i +δ o > 0 This means that activity (lack of passivity) of one system can be compensated by passivity of the other. 15 / 17

Passivity of interconnections We only mention these as facts, without proofs (see references). 1. A parallel interconnection (feedforward) of two passive systems remains passive. 2. A negative feedback interconnection of two passive systems remains passive. 3. In the feedback configuration of the previous slides with u 2 = 0, if one of the systems is passive and the other strictly input passive, the feedback combination is L 2 -stable. 16 / 17

Examples and notes on loop transformations With causal weights, techniques similar to those used in the SGT can be applied. Non-causal weights require the more advanced multiplier theory (Desoer and Vidyasagar). Linear invertible shifts can be used as in the SGT. A very general analysis of system interconnections involving active and passive systems has been developed recently by prof. Antsaklis and his group. See for instance: Zhu, Xia and Antsaklis, Passivity analysis and passivation of feedback systems using passivity indices, 2014 American Control Conference. 17 / 17