15.07.2014 Views

MECH550F: Multivariable Feedback Control Today's topics Why ...

MECH550F: Multivariable Feedback Control Today's topics Why ...

MECH550F: Multivariable Feedback Control Today's topics Why ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

<strong>MECH550F</strong>: <strong>Multivariable</strong><br />

<strong>Feedback</strong> <strong>Control</strong><br />

Lecture 2<br />

Internal stability of feedback systems<br />

Robust controller design process<br />

(Review)<br />

Reference<br />

<strong>Control</strong>ler<br />

Actuator<br />

Input<br />

Sensor<br />

Disturbance<br />

Plant<br />

Output<br />

Dr. Ryozo Nagamune<br />

Department of Mechanical Engineering<br />

University of British Columbia<br />

4. Implementation<br />

<strong>Control</strong>ler<br />

3. Design<br />

Uncertain model<br />

2. Analysis<br />

1. Modeling<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 1<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 2<br />

Today’s s <strong>topics</strong><br />

<strong>Why</strong> feedback?<br />

• Nominal stability of the feedback system<br />

C(s)<br />

P(s)<br />

• <strong>Why</strong> feedback?<br />

• What have you learned so far about nominal stability?<br />

• Well-posedness<br />

of the feedback system<br />

• Internal stability of the feedback system<br />

• Stabilizing controllers<br />

• To achieve system performance under<br />

uncertainties<br />

• Modeling error<br />

• System variations<br />

• Nonlinearities<br />

• Unknown signals<br />

• Stabilization of an unstable system<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 3<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 4<br />

1


Issue in stabilization<br />

by series connection<br />

• Consider three stable open-loop systems:<br />

• Pole/zero cancellation in unstable region by series<br />

connection is VERY BAD!<br />

• Unstable system MUST be stabilized by feedback!!!<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 5<br />

Issue in stabilization<br />

by series connection (cont’d)<br />

• <strong>Why</strong> is unstable cancellation by series<br />

connection not allowed?<br />

• Plant model is never exact! (and hence cancellation<br />

will not occur in real world.)<br />

• Neither nonzero initial condition nor input disturbance<br />

is allowed. (very fragile)<br />

• Some internal signal may go unbounded, even if<br />

output is bounded.<br />

• Internal stability (that we define later) can detect<br />

such cancellation!<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 6<br />

Today’s s <strong>topics</strong><br />

Nyquist stability criterion (Review)<br />

• Nominal stability of the feedback system<br />

C(s)<br />

P(s)<br />

• <strong>Why</strong> feedback?<br />

• What have you learned so far about nominal stability?<br />

• Well-posedness<br />

of the feedback system<br />

• Internal stability of the feedback system<br />

• Stabilizing controllers<br />

• P: # OL poles in open RHP<br />

(given)<br />

• N: # of clockwise<br />

encirclement of -11 by<br />

Nyquist plot of OL transfer<br />

function L(s) ) (counted by<br />

using Nyquist plot of L(s))<br />

15<br />

10<br />

5<br />

0<br />

-5<br />

-10<br />

-15<br />

-5 0 5 10 15<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 7<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 8<br />

2


Observer-based based control (review)<br />

Comments<br />

Plant<br />

Observer<br />

Closed-loop loop A-matrixA<br />

• Nyquist stability criterion<br />

• Graphical method to decide NS<br />

of feedback (FB) system in s-<br />

domain.<br />

• An assumption: No hidden<br />

unstable pole/zero cancellation.<br />

• Algebraic method to decide<br />

NS of FB system in s-domain,<br />

even with hidden unstable<br />

pole/zero cancellation?<br />

C(s)<br />

P(s)<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 9<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 10<br />

Today’s s <strong>topics</strong><br />

Some terminologies<br />

• Nominal stability of the feedback system<br />

• Real rational transfer function<br />

YES<br />

NO<br />

C(s)<br />

P(s)<br />

• Proper transfer function<br />

• <strong>Why</strong> feedback?<br />

• What have you learned so far about nominal stability?<br />

• Well-posedness<br />

of the feedback system<br />

• Internal stability of the feedback system<br />

• Stabilizing controllers<br />

• Strictly proper transfer function<br />

• Biproper: : Inverse is also proper<br />

• Real rational proper transfer matrix<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 11<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 12<br />

3


A feedback structure<br />

Well-posedness<br />

posedness: : Definition<br />

P1 & P2: P : real rational proper<br />

• A feedback system is called well-posed<br />

if all closed-loop loop transfer matrices are<br />

• well-defined (inverses exist) and<br />

• proper.<br />

(The argument “s” is omitted.)<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 13<br />

Remark: Any meaningful feedback system has<br />

to be well-posed!<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 14<br />

Physical meanings<br />

Examples<br />

• If not well-defined, for an external signal, the<br />

internal signal is not uniquely determined.<br />

• If not proper, for high frequency signals, the gain<br />

goes to infinity. (Physically unrealizable!)<br />

• Roughly speaking, not-well<br />

well-posed feedback<br />

system does not make sense.<br />

Ex.1<br />

Ex.2<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 15<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 16<br />

4


Examples (cont’d)<br />

Well-posedness<br />

posedness: : Condition<br />

Ex.3<br />

• The feedback system<br />

is well-posed<br />

if and only if<br />

Remark:<br />

Non-proper!<br />

because<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 17<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 18<br />

Well-posedness<br />

condition in SS<br />

Examples (cont’d)<br />

Notation:<br />

• The feedback system is well-posed if and only if<br />

Ex.4<br />

<strong>Feedback</strong> system is NOT well-posed!<br />

• A sufficient condition (We often use this!)<br />

If one of P1(s) P (s) or P2(s) P<br />

is strictly proper (i.e. D1=0 D<br />

or<br />

D2=0), =0), the feedback system is well-posed.<br />

Ex.5<br />

<strong>Feedback</strong> system is well-posed!<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 19<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 20<br />

5


Today’s s <strong>topics</strong><br />

Internal stability: Definition<br />

• Nominal stability of the feedback system<br />

C(s)<br />

P(s)<br />

• <strong>Why</strong> feedback?<br />

• What have you learned so far about nominal stability?<br />

• Well-posedness<br />

of the feedback system<br />

• Internal stability of the feedback system<br />

• Stabilizing controllers<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 21<br />

• Assume that the CL<br />

system is well-posed.<br />

• The CL system is called<br />

internally stable if<br />

All the poles are in the open LHP.<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 22<br />

Example<br />

Remarks<br />

CL system is well-posed! <strong>Why</strong>?<br />

CL system is NOT internally stable!<br />

• Internal stability is a basic requirement for a<br />

practical feedback system.<br />

• Internal stability guarantees that all signals in the<br />

CL system are bounded, provided that the<br />

exogenous signals (w) are bounded.<br />

• Exogenous signals (w) are models of, for<br />

example, initial conditions, noises, and modeling<br />

errors.<br />

• Roughly speaking, not-internally<br />

internally-stable feedback<br />

system does not work fine.<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 23<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 24<br />

6


Today’s s <strong>topics</strong><br />

Stabilizing controllers for a stable plant<br />

• Nominal stability of the feedback system<br />

C(s)<br />

P(s)<br />

• All stabilizing negative FB controllers C(s) ) for a<br />

stable plant P(s) ) is parameterized as<br />

• <strong>Why</strong> feedback?<br />

• What have you learned so far about nominal stability?<br />

• Well-posedness<br />

of the feedback system<br />

• Internal stability of the feedback system<br />

• Stabilizing controllers<br />

• Q(s) ) is any stable proper rational transfer function<br />

matrix, i.e.,<br />

• Youla parameterization (Q-parameterization)<br />

• A class of all the stabilizing controllers (even for an<br />

unstable plant: omitted in the lecture)<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 25<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 26<br />

Example<br />

Comments<br />

• Plant<br />

• All the stabilizing controllers<br />

• How to select Q?<br />

• Ex. If we want C to have an integrator,<br />

• Advantages of Youla parameterization<br />

• For a stable plant, the structure of the controller is<br />

identical to the one for (well-studied classical!)<br />

Internal Model <strong>Control</strong> (IMC).<br />

• Search over all stabilizing controllers is replaced by a<br />

search over stable Q(s).<br />

• All CL transfer functions are linear (or affine) in Q,<br />

simplifying optimization problems.<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 27<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 28<br />

7


Internal model control (IMC)<br />

Summary<br />

Q(s)<br />

Real plant<br />

P(s)<br />

• <strong>Feedback</strong> only the new information.<br />

• Design of Q(s) ) is relatively easy. (Proper and<br />

stable approximation of the inverse of P(s))<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 29<br />

• Well-posedness<br />

• Internal stability in s-domains<br />

• Stabilizing controllers<br />

• Announcements<br />

• No lecture on Sept 22 (Tue).<br />

• Today’s s HW is due Sept 22 (Tue) 5pm.<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 30<br />

Homework<br />

Homework (cont’d)<br />

• HW1: Prove<br />

• HW2: Check the well-posedness<br />

posedness, , as well as internal<br />

stability (if well-posed), for the following feedback<br />

systems. If unstable, stabilize the feedback system by<br />

modifying P2 P (P1 is fixed).<br />

•<br />

•<br />

• HW3: For MIMO feedback system (stable P(s)),<br />

verify that the following transfer functions are<br />

affine in Q(s) ) by using Youla parameterization.<br />

• From r to e<br />

• From r to u<br />

• From r to y<br />

• From d to e<br />

• From d to u<br />

• From d to y<br />

C(s)<br />

P(s)<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 31<br />

2009/10 T1 <strong>MECH550F</strong> : <strong>Multivariable</strong> <strong>Feedback</strong> <strong>Control</strong> 32<br />

8

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!