forked from bmnn/evrythg
-
Notifications
You must be signed in to change notification settings - Fork 0
/
APR_L5.txt
940 lines (597 loc) · 22.3 KB
/
APR_L5.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
Passivity
Ott Christian
Institute of Robotics and Mechatronics
German Aerospace Center (DLR)
http://www.robotic.dlr.de/Christian.Ott
These slides are based on [1] and the (German) lecture notes of Univ. Prof. Andreas Kugi,
Chair of System Theory and Automatic Control, Saarland University.
Overview
Introduction
The concepts of dissipative systems and passivity
Properties of passive systems
The relation between passivity and Lyapunov stability
Euler‐Lagrange systems and Hamiltonian systems
Linear passive systems & positive real systems
Relation to adaptive control
Literature
Introduction
Loosely speaking, the concept of passivity allows to generalize the principle of energy
conservation of closed physical systems: For a closed physical system the change of
energy is zero, i.e. energy cannot be generated or annihilated by the system itself.
The theoretical concept of passivity, however, is more general and must not be always
related to physical energy.
Still this analogy is very useful for understanding the theoretical concepts.
Consequently, we will start with a simple mechatronical system which shall serve as an
example plant throughout this lecture.
Simple electro‐magnetic Valve (1/2)
Mounting
Coil with N windings
Fext
Model Assumptions:
• Magnetic resistance of the mounting and the moving part are neglected.
• Geometric assumptions: h D
h
• The magnetic permeability of the air and of the sliding gap are given by
0 4 107 VsA 1m 1
Simple electro‐magnetic Valve (2/2)
V pin pout pdiss
1 L( z ) 2
Vmag
2 z
z
Vmag L( z ) I 2
0 N 2 D 2 ( D )b
L( z )
4(h z )( D )b D 2
Fmag
zv
v Fmag cz dv Fext
L( z )
1
Iv
I
U RI
z
L( z )
x ( z , v, I )
u (U , Fext )
Fext
V
L( z ) I 2 mv 2 cz 2
V UI Fext v dv 2 RI 2
pin pout
pdiss
V ( x(t )) V ( x(0)) s (U , Fext , I , v)d
s (U , Fext , I , v) UI Fext v
Dissipative Systems (1/3)
Generic time‐invariant nonlinear system: x f ( x, u )
y h ( x, u )
x X n
u U m
y Y p
Based on the input and output, we can introduce the so‐called supply rate
s (u , y ) : U Y ,
which is a real‐valued function.
Definition: The system (1) is said to be dissipative with respect to the supply rate
s(u,y), if there exists a non‐negative function
S ( x) : X ,
called the storage function, such that the so‐called dissipation inequality
S ( x) S ( x(0)) s (u ( ), y ( ))d
is fulfilled for all initial values x(0), all input signals u(t) , and all times t 0. If
S ( x) S ( x(0)) s (u ( ), y ( ))d
holds, the system is said to be lossless with respect to s.
Dissipative Systems (2/3)
Interpretation: The dissipation inequality expresses the fact that the “energy” S of the
system at any time is at most equal to the sum of the initial “energy” S(x(0)) and the
supplied external “energy” 0 s (u ( ), y ( ))d .
How much “energy” can we extract from the system via s?
The available storage is given by S a ( x(0)) : sup
s (u ( ), y ( ))d .
0
uU
t 0
Theorem: The system (1) is dissipative with respect to s if and only if the available
storage is finite S a ( x (0)) x(0) X .
In that case the available storage is a lower bound for all possible storage functions S,
and the inequality
0 S a ( x) S ( x)
holds.
Proof: see [1].
x X
Dissipative Systems (3/3)
If the storage function is continuously differentiable with respect to x, then the
dissipation inequality
S ( x) S ( x(0)) s (u ( ), y ( ))d
is equivalent to the differential dissipation inequality
S ( x) s (u (t ), y (t )), t 0
which often is much easier to check.
Remark: If a function S(x) fulfills the dissipation inequality and is bounded from below
by some finite negative value
S ( x) ,
then the non‐negative function S(x)‐ can be used as a storage function. Thus, it is
sufficient to show that S(x) is bounded from below.
Example: The electro‐magnetic valve is dissipative with respect to the supply rate
s (U , Fext , I , v) UI Fext v
where the inputs and outputs are assigned as u (U , Fext ),
y ( I , v)
Passivity (1/2)
Passivity can be seen as a special case of dissipativity. Let us again consider a general
time‐invariant nonlinear system of the form
x f ( x, u )
y h ( x, u )
x X n
( 2)
u U m
y Y m
but now with the number of inputs and outputs equal, i.e. m=p.
Definition: The system (2) is said to be passive if it is dissipative with respect to the
supply rate
s (u , y ) y T u.
It is strictly input passive if there exists a >0 such that it is dissipative with respect to
s (u , y ) y T u || u ||2
It is strictly output passive if there exists a >0 such that it is dissipative with respect to
s (u , y ) y T u || y ||2
A lossless passive system is also called conservative.
Passivity (2/2)
Physical Interpretation: If the product represents a physical power and S(x) is the
yT u
physical energy stored in the system, then the passivity inequality
S ( x) S ( x(0)) y T ud
means that the increase of the system’s energy is smaller or equal than the energy
supplied via the input.
Example: The electro‐magnetic valve is passive with respect to the inputs and
outputs
u (U , Fext ), y ( I , v)
Moreover, from
pdiss dv 2 RI 2
one can see that it is even strictly output passive.
Examples of passive systems:
• Integrator with the state space representation x u , y x
• Nonlinear mass‐spring‐damper like systems of the form
m d ( x) x k ( x) u, k ( x) V ( x) / x
y x, d ( x) 0, V ( x) 0x
Properties of Passive Systems (1/3)
Interconnection Properties of Passive Systems:
A)The parallel interconnection of passive systems results in a passive system again.
B)The feedback interconnection of passive systems results in a passive system again.
C)The series interconnection of two passive systems in general is not passive. If the
connecting element, however, is “energy conserving” (see figure below), i.e. if it
fulfills the condition
y
u I y2 y I d 0,
then this series interconnection is also passive. Remark: This property indeed is
utilized in some passivity based controller design methods.
S1 ( x1 ) S1 ( x1 (0)) y1 u1d
A) Parallel Interconnection
u1 u
u2 u
Passive
System 1
Passive
System 2
y y1 y2
S 2 ( x2 ) S 2 ( x2 (0)) y2 u2 d
S1 ( x1 ) S 2 ( x2 ) S1 ( x1 (0) S 2 ( x2 (0)) ( y1 y2 ) ud
0
S ( x1 , x 2 )
Properties of Passive Systems (2/3)
e1 , e2
B) Feedback Interconnection, input: , output: y1 , y2
Passive
System 1
S1 ( x1 ) S1 ( x1 (0)) y1 u1d
S 2 ( x2 ) S 2 ( x2 (0)) y2 u2 d
u1 y2 e1
u 2 y1 e2
Passive
System 2
S1 ( x1 ) y1 u1 y1 y2 y1 e1
S ( x ) yT u yT y yT e
S1 ( x1 ) S 2 ( x2 ) y1 e1 y2 e2
Properties of Passive Systems (2/3)
C) Power conserving interconnection, in which the connecting element fulfills the condition
y u
Passive
System 1
u1 u I e1
u 2 y I e2
y2 y I d 0,
Power‐
conserving
interconnection
Passive
System 2
S1 ( x1 ) y1 u1 y1 u I y1 e1
S ( x ) yT u yT y yT e
S1 ( x1 ) S 2 ( x2 ) ( y1 u I y2 y I ) y1 e1 y2 e2
() 0
Typically, system 1 can be a passive plant and system 2 a passive controller. As an
connecting element one can choose a system of the form
u I 0
y C ( x )T
I
C ( x) y1
where C(x) can be freely chosen so far.
0 y2
Lyapunov Stability: A Short Review
Consider the autonomous system x f (x)
x X n
Definition: An equilibrium point x=0 of (*) is said to be …
• stable, if for each >0 there exists a () such that
|| x(0) || || x(t ) || t 0
• attractive, if there exists an r>0 such that
|| x(0) || r lim || x(t ) || 0
t
• asymptotically stable, if it is stable and attractive.
• globally asymptotically stable, if it is asymptotically stable and its region of
attraction is the whole state space.
Theorem: Let x=0 be an equilibrium point of (*). Let V(x) be a continuously
differentiable positive definite function V (0) 0, V ( x) 0, x 0
which fulfills
V ( x)
V ( x)
f ( x) 0, x X
x
then the equilibrium point is stable. If moreover V ( x) 0, x X
then it is asymptotically stable.
Passivity & Lyapunov Stability (1/2)
x X n
x f ( x, u )
y h ( x, u )
( 2)
u U m
y Y m
Theorem: Suppose the system (2) is passive with a continuously differentiable storage
function S(x) and assume that the point x=0 is an equilibrium of the unforced system
(2) with u=0. If the function S(x) is positive definite at the point x=0, then this point is a
stable equilibrium of the unforced system
x f (x,0),
and the function S(x)‐S(0) is a Lyapunov function.
Remark: For asymptotical stability one may refer to La Salle’s invariance principle.
Two properties which are often used in the literature and which are useful for proving
asymptotic stability in case of strict output passivity are given in the following definition.
Definitions:
lim x 0
• If for the system (2) u=0 and y=0 for all t>0 implies then the system is said
t
to be zero‐state detectable.
• If for the system (2) u=0 and y=0 for all t>0 implies x=0 for all t>0 then the system is
said to be zero‐state observable
Passivity & Lyapunov Stability (2/2)
x f ( x, u )
y h( x )
x X n
u U m
y Y m
A simple controller: Consider the above system with the storage function S(x).
S ( x) y T u
If the system is passive but not strict output passive, then one obvious way to make the
system strict output passive is to apply a static output feedback
u ky v, k 0
with v as a new input. Then the closed loop system satisfies
S ( x) y T v k || y ||2 .
Stability properties:
Assume that the storage function is positive definite and that the system is zero‐state
u ky
detectable, then the feedback asymptotically stabilizes the system.
Euler Lagrange Systems (1/6)
Let us consider a Lagrangian system with generalized coordinates q (q )
and generalized velocities v q.
For simple Lagrangian systems, the Lagrange function L(q,v) is given by the difference
of kinetic and potential energy
L(q, v) T (q, v) V (q ).
The equations of motion can be computed by the Euler‐Lagrange equations
d L( q, v) L( q, v)
Qk
dt v k
q k
k 1,..., n.
Let us assume that the Lagrangian function is non‐degenerated, i.e. the matrix
2
i j L ( q, v )
v v
is regular (non‐singular).
Example (Robot Manipulator): T (q, v)
v M (q )v.
• The inertia matrix M(q) is symmetric and positive definite
• The potential function V(q) corresponds to the gravity potential.
Euler Lagrange Systems (2/6)
Using the generalized impulse coordinates
L(q, v)
p
v k
k 1,..., n.
and the Legendre‐Transformation
(q k , v k ) (q k , p k )
one can compute the Hamiltonian equations
d k H (q, p )
q
p k
H (q, p )
p
Qk ,
q k
k 1,..., n.
with the Hamilton function
H (q, p ) p k v k L(q, v).
k 1
Euler Lagrange Systems (3/6)
H (q, p ) p k v k L(q, v).
k 1
Example (Robot Manipulator):
T ( q, v )
v M (q )v.
L(q, v) T (q, v) V (q ) v M (q )v V (q ).
p
L(q, v)
M ( q )v
v
1
H (q, p ) vT M (q )v vT M (q )v V (q ) .
2
v T M ( q )v V ( q )
For the robot manipulator the Hamiltonian function corresponds to the robots energy.
Euler Lagrange Systems (4/6)
On the relation between the Euler‐Lagrange equations and the Hamiltonian equations:
H (q, p ) p k v k L(q, v).
k 1
p
n
j
L(q, v) v
H (q, p )
v
vk qk
v p j k
j 1 p
v j p k
p k
n
H (q, p )
v j L(q, v) v j L(q, p )
d L(q, v)
Qk
pj k
Q p
q k
q k
dt v k
j 1 q v q k
d L(q, v) L(q, v)
Qk
dt v k
q k
Euler Lagrange Systems (5/6)
Passivity considerations:
d k H (q, p )
q
p k
H (q, p )
Qk ,
p
q k
k 1,..., n.
Time derivative of the Hamilton function:
H (q, p )
H (q, p )
q
H ( q, p )
q
p
H (q, p ) H (q, p ) H (q, p ) H (q, p ) H (q, p )
p
q
p
p
q
H (q, p )
Q vT Q
p
From this one can see that the Euler‐Lagrange system is passive with respect to the
input Q and the output v with the H(q,p) as the corresponding storage function. In this
context it is also said that v is the collocated output with respect to the input Q.
Euler Lagrange Systems (6/6)
Let us assume that the generalized forces Q consist of some external control inputs
and some dissipative forces Qd R (v)
v
Q Qe Qd .
The dissipative forces are described by the Rayleigh’s dissipation function R(v) which
fulfills
v
R ( v ) v 0.
Definition: If the Rayleight’s dissipation function fulfills the inequality
R (v )v k v k ,
v
k 1
k 0, k 1,..., n.
Then the Lagrangian system is said to be fully damped.
From this it can easily be verified that a fully damped Euler‐Lagrange system is output
passive with respect to the input e and the output v with the H(q,p) as the
corresponding storage function.
Linear Passive Systems (1/3)
Consider a time‐invariant linear SISO (single input single output) system in state space
form
x Ax bu
y c T x du
or equivalently its transfer function
G (s)
y(s)
1
c T sE A b d
u (s)
Theorem: The strictly stable linear SISO system (*) is
• passive if and only if ReG ( j ) 0
• input passive if and only if
0 : ReG ( j ) 0
• output passive if and only if
ReG ( j ) | G ( j ) |2 0
Notice that with this theorem passivity of linear SISO systems can easily be checked by
examining the Nyquist plot of the transfer function.
Linear Passive Systems (2/3)
yud 0
yu u d 0
yu y d 0
x(t ), u (t ), y (t ) L2 (, )
Fourier transformation: x ( )
x(t ) exp( jt )dt
ˆ ˆ
Parceval’s theorem: x (t ) y (t )dt x, y x, y
2
It is assumed that all signals are causal: u (t ) 0, y (t ) 0
u (t ) t T
Let us define the operator ()T : uT (t )
t T
0
x( ) y * ( ) d
t0
Linear Passive Systems (3/3)
u (t ) y (t ) dt uT (t ) y (t )dt
2
uT ( ) y * ( )dt
y ( ) G ( j )u ( )
2
2
2
u (t ) y (t )dt
G * ( j )u * ( )uT ( ) d
ReG ( j ) j ImG ( j ) | uT ( ) |2 d
ReG ( j ) | uT ( ) |2 d
Equivalence of the passivity definition and the statement of the theorem:
0 : ReG ( j ) 0
u (t ) y (t )dt u (t )dt
u (t ) y (t ) dt
2
2
Parceval
Re(G ( j )) | u
-
| uT ( ) | d u 2 (t ) dt
( ) |2 d 0
Positive Real Systems (1/4)
For linear systems instead of passivity the notion of positive real systems is often used.
Definition: A transfer function is positive real (PR) if and only if
ReG(s) 0
s : Re(s) 0
G( s )
0
If is positive real for a then G(s) is said to be strictly positive real (SPR).
Relation between passivity and positive real systems:
A linear SISO system is passive if and only if its transfer function is positive real.
Example:
G( s)
s
0
Using s j one gets
G( s)
j
j ( ) 2 2
from which one can see that the transfer function is strictly positive real.
Positive Real Systems (2/4)
Theorem: A transfer function G(s) is positive real if and only if
1. G(s) has no poles with Resi 0
j
2. ReG ( j ) 0 for all for which is not a pole of G(s)
si ji
3. all poles of the form are distinct. For finite i
the limit
lim ( s j )G ( s )
s i
i
must be positive and real. For infinite the limit
G ( j )
j
must be positive and real.
Some simple necessary conditions for SPR:
• G(s) must be strictly stable.
• The Nyquist plot of G(s) lies entirely in the right half complex plane.
• G(s) has relative degree 0 or 1.
• G(s) is strictly minimum‐phase (i.e. all zeros are strictly in the left half plane).
Positive Real Systems (3/4)
Let us look at the relation between stability and passivity in the linear case:
x Ax
Review: For any strictly stable linear system of the form one has:
Q Q T 0
V ( x)
x Px
P P T 0 : AT P PA Q
V ( x ) x T Px x T Px x T AT Px x T PAx x T Qx
Let us consider a linear system of the form
x Ax bu
y cT x
with a Hurwitz matrix A (i.e. the free system shall be strictly stable)
x Px
V ( x ) x T Pbu x T Qx
c Pb
V ( x ) yu
In case of one gets x T Qx and thus a passive system.
V ( x)
Notice: For a given system of the form (*) with fixed inputs (i.e. for fixed b) one thus
c Pb
can find an infinity of outputs via for which the linear system is passive!
Positive Real Systems (4/4)
Kalman‐Yakubovich Lemma: Consider a controllable linear time‐invariant system of
the form
x Ax bu
y cT x
The transfer function G ( s ) c T ( sE A) 1 b
is strictly positive real if and only if there exist symmetric and positive definite matrices
P and Q such that
AT P AP Q
Pb c
Many variations of this lemma (for PR and SPR) exist in the literature. Sometimes it is
also called the “positive real lemma”. It gives a connection between state space
representations, transfer functions, and positive realness of linear systems. Also,
extensions to MIMO systems can be found in the literature.
Relation to Adaptive Control
Lemma: Consider two signals and related to the dynamical system
e (t )
(t ) m
e(t ) G ( s )( (t )T (t ))
G(s)
(t ) n
where is a strictly positive real transfer function, and is a measurable
(t )
vector. If the vector varies according to
(t ) (t )e(t )
with a positive constant , then and are globally bounded. Furthermore,
e (t )
(t )
0
(t )
if is bounded, then
e(t ) 0 as t
Applications: Output feedback adaptive control
Derivation of modified parameter update laws (as passive mappings)
Example:
Literature
[1] Van der Schaft A., L2‐Gain and Passivity Techniques in Nonlinear Control,
Springer, London, 2nd Ed., 1999.
[2] Slotine E., Li W., Applied Nonlinear Control, Prentice Hall, 1991.
[3] Sepulchre R., Jancovic M., Kokotovic P., Constructive Nonlinear Control,
Springer, London, 1997.