Stability

Introduction to Control of port-Hamiltonian systems
-
Stabilization of PHS
-
Doctoral course, Université Franche-Comté, Besançon, France
Héctor Ramı́rez, Yann Le Gorrec and Yongxin Wu
AS2M, FEMTO-ST UMR CNRS 6174,

Université de Franche-Comté,
École Nationale Supérieure de Mécanique et des Microtechniques,
26 chemin de l’épitaphe, F-25030 Besançon, France.
-
22-24 March, 2017
Doctoral course UFC-ST 1 / 103

References
Relevant references:
1. Brogliato, B., Lozano, R., Maschke, B., and Egeland, O. (2007). Dissipative
Systems Analysis and Control. Communications and Control Engineering Series.
Springer Verlag, London, 2nd edition edition.
2. A.J. van der Schaft, L2-Gain and Passivity Techniques in Nonlinear Control, Lect.
Notes in Control and Information Sciences, Vol. 218, Springer-Verlag, Berlin,
1996, p. 168, 2nd revised and enlarged edition, Springer-Verlag, London, 2000
(Springer Communications and Control Engineering series), p. xvi+249.
3. Jeltsema. D, van Der Schaft, A. Memristive port-Hamiltonian systems,
Mathematical and Computer Modelling of Dynamical Systems - MATH COMPUT
MODEL DYNAM SYST 01/2010; 16(2). DOI:10.1080/13873951003690824
4. van der Schaft, A.J and Jeltsema, D. Port-Hamiltonian Systems: from Geometric
Network Modeling to Control, Module M13, HYCON-EECI Graduate School on
Control, April 07–10, 2009.

Outline
1. Stability: definitions
2. Stabilization of PHS
3. Control by interconnection
4. Interconnection and Damping Assignment – Passivity Based Control
5. Final remarks and future work

Some basic notions on stability
We are considering the following class of state space model
ẋ(t) = f (x(t)),
(1)
x(0) = x0 ,
with t ∈ R and x ∈ Rn . Furthermore f : Rn × Rm → Rn . It is assumed that f (x(t))

satisfies the standard assumptions for existence and uniqueness of solutions, i.e., that
f (x(t)) is Lipschitz continuous with respect to x, uniformly in t, and piecewise
continuous in t.
We want to analyse the dynamics of the system
• Do the solutions of (1) remain bounded in time?

• If yes, do they in addition converge to some equilibrium point?
• If yes, can we say anything of the speed of convergence?
• Can we modify the solution with some external control input to impose a desired
closed-loop behaviour?

We are considering the following class of state space model
ẋ(t) = f (x(t)),
(2)
x(0) = x0 ,
with t ∈ R and x ∈ Rn . Furthermore f : Rn × Rm → Rn . It is assumed that f (x(t))

satisfies the standard assumptions for existence and uniqueness of solutions, i.e., that
f (x(t)) is Lipschitz continuous with respect to x, uniformly in t, and piecewise
continuous in t.
Equilibrium position
An equilibrium point x ∗ is a solution to
ẋ(x ∗ ) = 0, i.e., f (x ∗ ) = 0
Without loss of generality we will assume that x ∗ = 0.

An equilibrium position x = 0 of system (1) is

1. Stable if for any > 0 and t0 ≥ 0, there exists a δ(, t0 ) > 0 such that kx0 k < δ
implies kx(t, x0 )k < for all t ≥ t0 ,
2. Uniformly Stable if δ does not depend on t0 ,
3. Asymptotically Stable if it is stable and for any t0 ≥ 0 there exists ∆(t0 ) > 0 such
that every solution x(t, x0 ) for which kx0 k < ∆ satisfies the relation
lim kx(t, x0 )k → 0 (3)

t→∞
4. Uniformly Asymptotically Stable if it is uniformly stable, ∆ does not depend on t0 ,

and (3) holds uniformly with respect to t0 and x0 in the domain t0 ≥ 0, kx0 k < ∆,

Some notions on stability
An equilibrium position x = 0 of system (1) is

1. Globally Asymptotically Stable if it is stable and relation (3) holds for any t0 ≥ 0
and x0 ,
2. Uniformly Globally Asymptotically Stable if it is uniformly stable and relation (3)
holds for any t0 ≥ 0 and x0 uniformly relative to t0 and x0 in the domain t0 ≥ 0,
x0 ∈ K , where K is arbitrary compact in the x-space,
3. Exponentially Asymptotically Stable if there exist positive constants ∆, M, and α
such that every solution x(t, x0 ), for which kx0 k < ∆, satisfies the relation
kx(t, x0)k < Mkx0 ke−α(t−t0 ) (4)
for all t ≥ t0 ≥ 0, and

4. Globally Exponentially Asymptotically Stable if there exist positive constants M
and α such that relation (4) holds for t ≥ t0 ≥ 0 and arbitrary t0 ≥ 0 and x0 .

Figures taken from:

http://www.math24.net/

How do we analyse stability?
Lyapunov stability theory
• Lyapunov’s direct method (second method) → non-linear systems

• Lyapunov’s indirect method (first method) → linear systems
Lyapunov’s direct method allows to determine the stability of a system without explicitly
integrating the differential equations. The method is a generalization of the idea that if
there is some “measure of energy” in a system, then we can study the rate of change
of the energy of the system to ascertain stability.

Let B be a ball of size around the origin, B = x ∈ Rn : x < kk.
Positive definite function

A continuous function V : Rn → R is a locally positive definite function if V (0) = 0 and
for x ∈ B , x 6= 0 → V (x) > 0. If B is the whole state space, then V (x) is globally
positive definite.
A positive definite function is like an energy function.

Lyapunov’s direct method
Theorem: Lyapunov stability Let V (x) be a non-negative function with continuous

partial derivatives such that
• V (x) is positive definite on B , and V̇ ≤ 0 locally in x and for all t, then the origin
of the system is locally stable (in the sense of Lyapunov).
• If in addition V (x) → ∞ when kxk → ∞, then the system is globally stable.
Theorem: Asymptotic stability Let V (x) be a non-negative function with continuous

partial derivatives such that
• V (x) is positive definite on B , and V̇ < 0, ∀x ∈ B /{0} and V (0) = 0 locally in x
and for all t, and then the origin of the system is locally asymptotically stable.
• If in addition V (x) → ∞ when kxk → ∞, then the system is globally
asymptotically stable.

Lyapunov’s direct method
Figure: taken from: http://www.math24.net/
For physical systems: relate the physical energy with Lyapunov functions

Stabilization of passive systems
Let us consider systems arising from some physical energy model. We then usually
have
Z t Z t
H(t) = H(t0 ) + u(τ )y(τ )dτ − S(x(τ ))dτ .
|0 {z } |0 {z }
supplied energy dissipated energy
So if H(x) qualifies as a Lyapunov function and S(x) vanishes at x = 0 (and only in
x = 0), then the system is asymptotically stable!
So why do we need the control then?

• What if we want to increase the rate of convergence?: damping injection,

• What if we want to stabilize at some different equilibrium point, x = x ∗ , x ∗ =
6 0:
Energy shaping,
• What if S(x) vanishes for some x 6= 0 or S(x) = 0?: damping injection + Energy
shaping

PHS are dissipative and passive systems
Consider the system
ẋ(t) = f (x(t), u(t)), y(t) = h(x(t), u(t)), (5)
with t ∈ R, x ∈ Rn , u ∈ Rm and y ∈ Rp . Furthermore f : Rn × Rm → Rn and

h : Rn × Rm → Rp . Let us addition define the supply rate w(t) = w(u(t), y(t)),
Z t
|w(u(τ ), y(τ ))dτ | < ∞
0
Dissipative systems
The system (5) is said to be dissipative if there exists a so-called storage function
V (x) ≥ 0 such that the following dissipation inequality holds:
Z t
V (x(t)) ≤ V (x(0)) + w(u(τ ), y(τ ))dτ
0
along all possible trajectories of (5) starting at x(0), for all x(0), t ≥ 0.

Passive systems
A particular case of dissipative systems are passive systems:
Passive systems
Suppose that the system (5) is dissipative with supply rate w(u, y) = u T y and storage
function V (x(t)) with V (0) = 0; i.e. for all t ≥ 0 we have that
Z t
V (x(t)) ≤ V (x(0)) + u(τ )> y(τ )dτ,
0
Then the system is passive.
Passive systems are a subclass of dissipative systems with the specific properties
• The supply rate is defined by the product between inputs and outputs
w(u, y) = u T y,
• The minimum of the storage function is at the origin V (0) = 0.

Passive systems
The dissipation of a passive system may also be explicitly taken into account:
Strictly passive systems

A system (5) is said to be strictly state passive if it is dissipative with supply rate
w = u > y and the storage function V (x(t)) with V (0) = 0, and there exists a positive
definite function S(x) such that for all t ≥ 0:
Z t Z t
V (x(t)) = V (x(0)) + u(τ )> y(τ )dτ − S(x(τ ))dτ ,
0 0
If the equality holds and S(x) = 0, then the system is said to be lossless
(conservative).
The function S(x) is called the the dissipation rate.

Passive systems
Why are we interested in passive systems?

• Many physical systems are passive with respect to the storage function defined by
their physical energy function and with respect to their natural supply rate (given
by the physical inputs and outputs),
• Its a non-linear approach (does not require any assumption of linearity),
• The physical energy may be used as a candidate Lyapunov function to analyse
stability.
• A “well defined” interconnection of passive system is again a passive system.

Stability of passive systems
Recall the passivity inequality

Z t
V (x(t)) ≤ V (x(0)) + u(τ )> y(τ )dτ,
0
if the storage function V is strictly increasing and x = 0 is an isolated global minimum,

then the output feedback u = 0 stabilizes the system at x = 0 (LaSalle’s inv. Principle).
u=0
V̇ ≤ 0
For many physical systems: Energy = storage function = Lyapunov function

Port-Hamiltonian control system
Within this formalism a physical system is defined by the interconnection between

energy storage elements, resistive elements, and the environment:
This defines a natural space: F := FS × FR × FP ; E := ES × ER × EP

Port-Hamiltonian control systems
The interconnection structure satisfies the power preserving property
es> fs + eR
>
fR + ep> fp = 0
or in terms of the energy storing elements
Ḣ(x(t)) = −es> fs = eR
>
fR + ep> fp
which yields the energy balance equation
Z t
>
H(x(t)) = H(x(0)) + eR (τ )fR (τ ) + ep> (τ )fp (τ )dτ
0
Dirac structure → port-Hamiltonian systems

The power preserving property is defined by the geometric notion of a Dirac structure,
which naturally defines port-Hamiltonian control systems.

Explicit PHS
∂H
ẋ = J − R + gu(t), f = J − R e + geext
∂x
∂H
y = g> = (g > e)
∂x
Two geometric structures: J(x) and R(x) = R(x)> ≥ 0, the interconnection and
damping matrices
The energy-balance is
∂H > ∂H
Ḣ = u > (t)y(x(t))− (x(t))R(x(t)) (x(t)),
∂x ∂x
Ḣ ≤ u > (t)y(t),
Dissipative PHS are strictly-passive if the Hamiltonian H is bounded from below.

Stability of strictly passive systems
Recall the strict passivity inequality

Z t Z t
V (x(t)) ≤ V (x(0)) + u(τ )> y(τ )dτ − S(x(τ ))dτ,
0 0
if the storage function V is strictly increasing, x = 0 is an isolated global minimum and

S(x(t)) only vanishes at x = 0, then the output feedback u = 0 asymptomatically
stabilizes the system at x = 0 (LaSalle’s inv. Principle).
u=0
V̇ ≤ −S(x(t))
For many physical systems: physical dissipation = dissipation rate

Examples: RLC circuit and MSD system
Show that the RLC and the MSD are strictly passive and give the corresponding PHS
models.

Example: RLC circuit
Let us consider a simple linear RLC circuit:
Constitutive relations
us = Vin
ur = RIr
φ = LIL
Q = CuC
Dynamic relations
dφ
Z t
uL = , or in integral form φ(t) = φ(t0 ) + uL (τ )dτ
dt 0
dQ
Z t
IC = , or in integral form Q(t) = Q(t0 ) + IC (τ )dτ
dt 0

P P
Interconnection relations (Kirchkoff’s laws): u = 0 voltage law, i = 0 current law
Using the interconnection relations together with the constitutive and dynamical
relations we obtain the state space model
dQ φ
=
dt L
dφ Q φ
= − − R + Vin
dt C L
with state variables x = [Q, φ] and input Vin .
• If the initial conditions Q(t0 ) and φ(t0 ) are known, together with the profile Vin ,
then the time evolution of the system is fully determined for all t > t0 .

What about the energy of the systems?

Energy = Energy stored in the capacitor + Energy stored in the inductor
1 φ2 1 Q2
H(x(t)) = +
2L 2C
The time variation of the energy is given by
2
∂H > dx

dH(x(t)) Q φ Q φ φ φ
= = − + Vin −R
dt ∂x dt C L C L L L
2
φ φ
= Vin −R = Vin IL − RIL2
L L
Hence, the balance equation characterizing the time variation of energy can be written
as
Z t Z t
H(t) = H(t0 ) + Vin (τ )IL (τ )dτ − RIL (τ )2 dτ
0 0

Example: mass-spring-damper system
Let us consider a simple linear translational MSD system:
Fs = Fin
FB = BvB
p = MvM
q = K −1 FK
Dynamic relations
dp
Z t
FM = , or in integral form p(t) = p(t0 ) + FM (τ )dτ
dt 0
dq
Z t
vK = , or in integral form q(t) = q(t0 ) + vK (τ )dτ
dt 0

Using the interconnection relations (Kirchkoff’s laws) together with the constitutive and
dynamical relations we obtain the state space model
dq p
=
dt M
dp q p
= − −1 − B + Fin
dt K M
with state variables x = [q, p] and input Fin .

Example: MSD system
What about the energy of the systems?

Energy = Energy stored in the mass + Energy stored in the spring
1 p2 1 q 2
H(x(t)) = +
2M 2 K −1
The time variation of the energy is given by
dH(x(t)) ∂H > dx q p q p p p 2

= = −1
− −1
+ Fin −D
dt ∂x dt K M K M M M
p p 2
2
= Fin −B = Fin vM − BvM
M M
The balance equation characterizing the time variation of energy can be written as
Z t Z t
H(t) = H(t0 ) + Fin (τ )vM (τ )dτ − BvM (τ )2 dτ
0 0

Port-modelling of physical systems
Let us look closer to the energy balance

Z t Z t
H(t) = H(t0 ) + uin (τ )y(τ )dτ − R(x)y(τ )2 dτ
0 0
| {z } | {z }
The balance equations expresses conservation of some physical quantity: Energy,

mass, volume, etc...
The existence of balance equations is the base for dissipative and passive system
theory.

Example: the RLC circuit
Let us first consider a lossless LC circuit. The energy is
1 Q2 1 φ2
H(x(t)) = +
2C 2L
The interconnection structure just characterize the exchange of energy between the
inductor and the capacitor:
0 1
J= .
−1 0
The internal dynamics of the system is then given by
" # " #
Q φ
∂H C L
ẋ = J =J φ =
∂x L
−QC

Let us consider the complete RLC circuit, with dissipation and input port. The energy
remains the same
1 Q2 1 φ2
H(x(t)) = +
2C 2L
inductor and the capacitor, but in this case we have to add an additional structure
matrix that characterizes the dissipation of the system and an input vector field

0 1 0 0 0
J= , R= , gu = u
−1 0 0 R 1
The complete dynamics of the system is now given by

" # " #
Q φ
∂H
ẋ = (J − R) + gu = (J − R) C φ + gu = L
∂x L
−Q
C
− R φL + Vin

Example: the MSD system
Let us first consider a lossless MS system. The energy is
1 q 2 1 p2
H(x(t)) = +
2 K −1 2M
mass and the spring:
0 1
J= .
−1 0
The internal dynamics of the system is then given by
q p
∂H K −1 M
ẋ = J =J p = q
∂x M
− K −1

Example: the MSD system
Let us consider the complete MSD system, with dissipation and input port. The energy
remains the same
1 q 2 1 p2
H(x(t)) = −1
+
2K 2M
The interconnection structure remains the same, but in this case we have to add an
additional structure matrix that characterizes the dissipation of the system and an input
vector field
0 1 0 0 0
J= , R= , gu = u
−1 0 0 B 1
The complete dynamics of the system is now given by
q p
∂H K −1 M
ẋ = (J − R) + gu = (J − R) p + gu = q p
∂x M
− K −1 − BM + Fin

1 φ2 1 Q2
H(x(t)) = + ≥ 0, H(0) = 0.
2L 2C
Hence H qualifies as a potential storage function. Now,
Z t Z t
H(t) = H(t0 ) + Vin (τ )IL (τ )dτ − RIL (τ )2 dτ .
0 0
The system is passive if we choose u = Vin and y = IL :

Z t
H(t) ≤ H(t0 ) + Vin (τ )IL (τ )dτ .
0
Furthermore, if the we choose the dissipation rate as S(x) = RIL (τ )2 , then the system
is strictly passive
Z t Z t
|0 {z } |0 {z }

1 q 2 1 p2
H(x(t)) = + ≥ 0, H(0) = 0.
2 K −1 2M
Hence H qualifies as a potential storage function. Now,
Z t Z t
H(t) = H(t0 ) + Fin (τ )vM (τ )dτ − BvM (τ )2 dτ .
0 0
The system is passive if we choose u = Fin and y = vM :

Z t
H(t) ≤ H(t0 ) + Fin (τ )vM (τ )dτ .
0
Furthermore, if the we choose the dissipation rate as S(x) = BvM (τ )2 , then the
system is strictly passive
Z t Z t
|0 {z } |0 {z }

Some remarks
• The chosen inputs and outputs correspond to the physical input and outputs of the
system: input voltage and input force / current in the inductor and velocity of the
mass
• If we eliminate the resistive components, resistor (R) and damper (B), the supply
rate is zero and the system is a lossless (conservative) passive system. Indeed,
Z t
H(t) = H(t0 ) + u(τ )y(τ )dτ
|0 {z }
supplied energy
i.e., the energy is conserved.

• The product u > y has the units of power, i.e., it defines a power product. This has
strong implications for modelling: if the input and outputs define power products
the power preserving interconnection of physical (passive) systems defines again
a physical (passive) system.


• What if we want to increase the rate of convergence?: damping injection,

• What if we want to stabilize at some different equilibrium point, x = x ∗ , x ∗ =
6 0:
Energy shaping,
• What if S(x) vanishes for some x 6= 0 or S(x) = 0?: damping injection + Energy
shaping

Stabilization of PHS: Damping injection
Consider the energy balance equation of a passive system:

Z t Z t
|0 {z } |0 {z }
And assume that H(x) qualifies as a Lyapunov function candidate. If we select the
input u = −Ky, with K a positive definite constant matrix, then the energy balance
equation becomes:
Z t Z t
H(t) = H(t0 ) −K y 2 (τ )dτ − S(x(τ ))dτ ,
0
| {z } |0 {z }
controller dissipated energy
Z t
H(t) = H(t0 ) − Ky 2 (τ )dτ + S(x(τ )) dτ .
|0 {z }
dissipated energy

Let us consider a simple linear translational MSD system:
Fs = Fin
FB = BvB
p = MvM
q = K −1 FK
Dynamic relations
dp
Z t
FM = , or in integral form p(t) = p(t0 ) + FM (τ )dτ
dt 0
dq
Z t
vK = , or in integral form q(t) = q(t0 ) + vK (τ )dτ
dt 0

Examples: the MSD system
1 q 2 1 p2
H(x(t)) = −1
+ ≥ 0, H(0) = 0.
2K 2M
Hence H qualifies as a storage function and as a candidate Lyapunov function. Now,
Z t Z t
H(t) = H(t0 ) + Fin (τ )vM (τ )dτ − BvM (τ )2 dτ .
0 0
The system is passive if we choose u = Fin and y = vM , and furthermore, if we select

u = −Ky , (Fin = −KvM ), then
Z t
2 2
H(t) = H(t0 ) − KvM (τ ) + BvM (τ ) dτ .
0
Z t
2
= H(t0 ) − (K + B) vM (τ )dτ
0 | {z }
B0
We have changed (increased) the system’s natural damping.

Stabilization of PHS: Energy Shaping
Consider the energy balance equation of a passive system:

Z t Z t
H(t) − H(t0 ) = u(τ )y(τ )dτ − S(x(τ ))dτ .
|0 |0
| {z }
stored energy {z } {z }
Assume that we want to change the closed-loop equilibrium to some forced (controlled)
equilibrium x = x ∗ . In that case H(x ∗ ) 6= 0, hence H(x) can no longer be used as
Lyapunov function!
We need to consider a new Lyapunov function candidate

Stabilization of PHS: Energy shaping
Let us consider the energy balance equation and assume we have no dissipation
Z t
H(t) − H(t0 ) = u(τ )y(τ )dτ
0
| {z }
controller
The idea is to construct a new energy function, by using the (state) feedback u = β(x)
Z t
Hd (x, x ∗ ) = H(x) − β(x(τ ))y(τ )dτ
0
such that Hd , with Hd (x ∗ ) = 0 qualifies as a Lyapunov function for the closed-loop

system.

If this function exist (...why should it exist?) it will be a state function such that
Hd (x) = H(x) + Ha (x), hence
Z t
Ha (x, x ∗ ) = − β(x(τ ))y(τ )dτ
0
The time derivative of Hd (x) along the trajectories of the system is given by
∂Ha >
Ḣd = Ḣ + Ḣa = Ḣ + ẋ
∂x
∂Ha >
⇒ ẋ = −β(x)y
∂x
Hence, for dynamical systems of the form ẋ = f (x, u), y = h(x), in order for the
function Ha to exist, the following PDE should be satisfied
∂Ha >
f (x, β(x)) = −β(x)h(x)
∂x

Some remarks
• Energy shaping requires the solution of a PDE: the matching equation. Not an
easy task for general non-linear systems
Z t
Ha (x, x ∗ ) = − β(x(τ ))y(τ )dτ
0
• The existence of solutions for the PDE is strongly related with the existence of
physical invariants. In the case of port-Hamiltonian systems: Casimir functions.
• For systems arising from physical applications the energy shaping technique has
been proven to be a powerful stabilization method.

us = Vin
ur = RIr
φ = LIL
Q = CuC
Dynamic relations
dφ
Z t
dt 0
dQ
Z t
dt 0

The state space model
dQ φ
=
dt L
dφ Q φ
= − − R + Vin
dt C L
with state variables x = [Q, φ], output y = φL = xL2 and input Vin . The energy of the
system is given by
1 x1 2 1 x2 2
H(x) = +
2C 2 L
• If Vin = 0, the natural equilibrium is x = (0, 0). If on other hand Vin = V ∗ , the
forced equilibrium point is x = (x1∗ , 0), with x1∗ = CV ∗ .

∂Ha >
The matching condition ∂x f (x, β(x)) = −β(x)h(x)

∂Ha x2 ∂Ha x1 x2 x2
− −R − β(x) = − β(x)
∂x1 L ∂x2 C L L
Notice that the forced equilibrium corresponding to the x2 coordinate already is a

minimum of the physical energy H(x), hence we only need to shape the closed-loop
energy in the x1 coordinate. Hence
Ha = Ha (x1 )
and the matching equation becomes
∂Ha x2 x2
= − β(x)
∂x1 L L
Hence, the function Ha exists if the feedback is chosen as β(x) = − ∂H

∂x
a
1

Beautiful!
The matching equation (PDE) is automatically solved for any Ha = Ha (x1 ) provided
that the state feedback is of the form β(x) = − ∂H
∂x
a
.
1
• It only remains to select Ha (x1 ) such that Hd = H + Ha has a minimum at
x ∗ = (x1∗ , 0).
Recall that the open-loop energy function is
1 x1 2 1 x2 2
H(x) = +
2C 2 L
Hence if we chose
1 2 1 1
Ha (x1 ) = x − + x1 x1∗
2Ca 1 Ca C
The closed-loop energy function Hd = H + Ha
2
1 (x1 − x1∗ ) 1 x2 2
Hd (x) = +
2 (C + Ca ) 2 L

Hd (x, x ∗ ) has a minimum at x ∗ = (x1∗ , 0) if and only if Ca > −C
The resulting controller is

∂Ha 1 2 1 1
β(x) = − (x) = − x − + x1 x1∗
∂x1 2Ca 1 Ca C

Final remarks
• We have revised some concepts from passivity based control techniques:

Damping injection and Energy Shaping
• We have exploited the natural passivity of the system to design stabilizing
controllers
• Works well in many applications, but.... we did not see the use of physical
invariants, the dissipation obstacle,...
What remains for tomorrow

• Control by interconnection
• IDA-PBC
• Irreversible port-Hamiltonian systems...


Let us recall the state space model of a port-Hamiltonian control system
∂H
ẋ = J(x) − R(x) (x) + g(x)u,
∂x
∂H
y = g > (x) (x),
∂x
where where x ∈ Rn is the state vector, u ∈ Rm , m < n, is the control action,

H : Rn → R is the total stored energy, J(x) = −J(x)> is the n × n natural
interconnection matrix, R(x) = R(x)> ≥ 0 is the n × n damping matrix, g(x), is the
n × m input map and u, y ∈ Rm , are conjugated variables whose product has units of
power.
∂H > ∂H
Ḣ = u > y − R ,
∂x ∂x
Ḣ ≤ u > y,

Control by interconnection
A controlled system may be viewed as a plant system interconnected with a control

system exchanging energy
The interconnection is power continuous if
u > (t)y(t) + uc> (t)yc (t) = 0, ∀t

For instance: a negative feedback
The negative feedback defines the following relation
u = −yc
⇒ u > y + uc> yc = −yc> y + y > yc = 0
y = uc

Let us consider the feedback
Now u = v − yc and uc = y + vc . Let the plant and controller have state variables x
and ξ and energy functions H(x) and H(ξ). If the maps u → y and uc → yc are
passive,
Then the map (v , vc ) → (y , yc ) is passive with energy function Hd (x, ξ) = H(x) + H(ξ).

Assume that the plant and the controller are PHS


 ẋ =[J(x) − R(x)] ∂H (x) + g(x)u


Σ: ∂x x ∈X
∂H
 y =g > (x)

 (x)
∂x
∂Hc

 ξ̇ =[Jc (ξ) − Rc (ξ)] ∂ξ (ξ) + gc (ξ)uc


Σc : ξ ∈ Xc
∂H
 yc =gc> (ξ) c (ξ)


∂ξ
Booth are passive systems, so a power preserving interconnection, u = −yc , y = uc ,

yields a passive closed-loop system.

The closed-loop systems looks
 
 ∂Hd (x)
" #
−g(x)gc> (ξ)

ẋ J(x) R(x) 0  ∂x
= −

>  ∂Hd
ξ̇  gc (ξ)g (x) Jc (ξ) 0 Rc (ξ)  (ξ)
| {z } | {z } ∂ξ
Jcl (x,ξ) Rcl (x,ξ)
" ∂Hd #
y g(x) 0 ∂x
(x)
= ∂Hd
yc 0 gc (ξ) (ξ)
| {z } ∂ξ
gcl
With total energy function
Hd (x, ξ) = H(x) + Hc (ξ)
We may equivalently write the closed-loop system as
∂Hd ∂Hd
ẇ = (Jcl − Rcl ) , ycl = gcl>
∂w ∂w
with w = [x ξ].

So, what now?

We would like to get an energy function in terms of x only: Hd = Hd (x), so that we can
assign the minimum at a desired point and characterize it in terms of the plant system.
In order achieve this, we must restrict the dynamics to a submanifold of the (x, ξ)
space parametrized by x. This means that we are looking for a submanifold
ΩC = (x, ξ) : ξ = F (x) − C
which is dynamically invariant, i.e.,

!
∂Fi >
ẋ − ξ̇i =0
∂x
ξ=Fi (x)−C

Casimir functions
Let us look for structural invariants that relates each state of the controller with the
states of the plant:
Ci (x, ξi ) = Fi (x) − ξi
In order to relate all the states of the controller with the state of the plant we define
F (x) = [F1 (x), F2 (x), . . . , Fnc (x)], and define the following Casimir function
n
X n
X
C= (Fi (x) − ξi ) = Ci (x, ξi )
i=1 i=1
C is an invariant of the system, hence
∂C > ∂C >

∂Hcl
Ċ = ẇ = Jcl =0
∂w ∂w ∂w
But furthermore, C is a structural invariants, so it should be invariant with respect to the

structure of the system:
∂C >
Jcl = 0
∂w

Casimir functions
Let us look for structural invariants that relates each state of the controller with the
states of the plant:
Xn n
X
C= Ci (x, ξi ) = (Fi (x) − ξi )
i=1 i=1
we obtain the following matching condition

" ∂Hd #
h
∂F >
i J(x) − R(x) −g(x)gC> (ξ) ∂x
(x)
(x) −I ∂H =0
∂x gC (ξ)g > (x) JC (ξ) − RC (ξ) d
(ξ)
| {z } ∂ξ
Matching condition
• Only the term in blue is considered in the matching condition because we want the
Casimir functions to be structural invariants of the system: not depend on
Hd (x, ξ).

The condition for existence of Casimir functions for the closed loop system
i J(x) − R(x) −g(x)gC> (ξ)
h
∂F > =0
(x) −I
∂x gC (ξ)g > (x) JC (ξ) − RC (ξ)
may be written out as
Matching equations
∂F > ∂F
(x)J(x) (x) = Jc (ξ)
∂x ∂x
∂F
R(x) (x) = 0 Dissipation obstacle!
∂x
Rc (ξ) = 0
∂F >
(x)J(x) = gc (ξ)g > (x)
∂x

The closed-loop dynamic then takes the form

h i ∂H ∂Hc
ẋ = J(x) − R(x) (x) − g(x)gC> (ξ) (ξ)
∂x ∂ξ
Using the second and fourth M.C. we get

h i ∂H ∂F ∂Hc

ẋ = J(x) − R(x) (x)+ (x) (ξ)
∂x ∂x ∂ξ
Since ξ = F (x) − C, we use the chain-rule for differentiation to establish
∂F ∂Hc ∂Hc
(x) (ξ) = (F (x) − C)
∂x ∂ξ ∂x

Hence we obtain:
h i ∂H ∂Hc

ẋ = J(x) − R(x) (x)+ (F (x) − C)
∂x ∂x
Or equivalently
h i ∂H
d
ẋ = J(x) − R(x) (x)
∂x
With closed-loop energy Hd (x) = H(x) + Hc (F (x) − C).

Let us interpret the control in terms of energy balancing. Since Rc = 0, the energy
balance equation of the controller is
dHc
= uc> yc
dt
Hence, along any invariant submanifold ΩC , we have
dHd dH dHc dH
= + = −u > y (uc> yc = −u > y)
dt dt dt dt
and integrating (up to some constant) we obtain
Z t
Hd (t) = H(t) − u > (τ )y(τ )dτ
0
| {z }
Hc
Rt
We obtain the general M.E!: Hc (t) = − 0 u > (τ )y(τ )dτ .

us = Vin
ur = RIr
φ = LIL
Q = CuC
Dynamic relations
dφ
Z t
dt 0
dQ
Z t
dt 0

Let us consider a RLC circuit, with dissipation and input port. The energy is
1 Q2 1 φ2
H(x(t)) = +
2C 2L
The interconnection and dissipation matrix and input vector field are

0 1 0 0 0
J= , R= , gu = u
−1 0 0 R 1
and the dynamic is

" # " #
Q φ
∂H C L
ẋ = (J − R) + gu = (J − R) φ + gu =
∂x L
−Q
C
− R φL + Vin
φ x2
with state variables x = [Q, φ], output y = L
= L
and input u = Vin
• If Vin = 0, the natural equilibrium is x = (0, 0). If on other hand Vin = V ∗ , the
forced equilibrium point is x = (x1∗ , 0), with x1∗ = CV ∗ .

Let us synthesis the controller using the M.C.s.

• From physical considerations we know that we only need to shape the x1
coordinate: F is only one scalar function.
• From M.E.2 we obtain that ∂F = 0.
∂x2
• Then, from M.E.1. we obtain that Jc = 0, and from M.E.3 that Rc = 0.
• Finally from M.E.4 we have that ∂F = gc (ξ) and that ξ ∈ R.
∂x1
We would like to have
2
1 (x1 − x1∗ ) 1 x2 2

1 1 1
Hc (x1 ) = x2
2Ca 1
− Ca
+ C
x1 x1∗ , such that Hd (x) = +
2 (C + Ca ) 2 L
This is achieved if we select F (x1 ) = x1 and C = 0 on the invariant submanifold such

∂F
that ξ = x1 . The control system is then given by (using condition ∂x = gc (ξ))
1
ξ̇ = uc
∂Hc
yc =
∂ξ

The pendulum without damping
The dynamic equations
p
q̇ =
m
ṗ = −mg sin(q) + u
with state variables x = [p, q], with q the configuration and p the momentum.
Propose a PH model and design a stabilizing controller using the Casimir method and
a control system given by:
ξ̇ = uc
∂Hc
yc =
∂ξ

Example: pendulum (model)
Consider the pendulum without damping

p
q̇ =
m
ṗ = −mg sin q + u
[q, p]T
with state variables x = with q the configuration and p the momentum.
The port Hamiltonian model is:
∂H0
!
d q 0 1 ∂q 0
= ∂H + u
dt p −1 0 0 1
| {z } ∂p | {z }
J g
∂H
!
0
∂q p
y = (0 1) ∂H0 = m
∂p
1 2
with Hamiltonian :H0 (q, p) = mg(1 − cos q) + 2m
p

pendulum (extended Casimir functions)
Recall: we look for Casimir functions such as:
C (q, p, xc ) = F (q, p) − xc
that is functions F (q, p) satisfying:

∂F
!
0 ∂F ∂p
1. input vector field is Hamiltonian: g = = J (x) =
1 ∂x − ∂F
∂q
∂F t ∂F
2. the gradient of F is transversal to g: Lg F (x) = 0 = ∂x
g (x) = ∂p
hence a generating function is: F (q, p) = −q

pendulum (control)
For the controller design we choose a function HC (xc ) such that

Hd (x) = H0 (x) + Hc ◦ F has a minimum at the desired equilibrium x∗ = (x1∗ , 0).
The simplest choice is given by
1
HC (xc ) = mg(cos xc − 1) + (xc + x1∗ )2
2
The control is finally obtained is:
∂HC
u=− (xc ) |xc =−q = mg sin q − (q − x1∗ )
∂xc
which is the well-known “proportional plus gravity compensation control

Remarks
• The port-Hamiltonian structure provide important information for finding the
solutions of the control system,
• The control has physical interpretation in terms of interconnection and energy
balancing
• The Casimir method can be used to analyse new stability profiles of
interconnected systems


Interconnection and Damping Assignment – Passivity Based
Control
Consider the following parallel RLC circuit
with x = [qC , φL ], the charge in

the capacitor and the flux in the
inductance
The PH model is
1 " ∂H #
with

q̇C −R 1 ∂qC 0
= ∂H + u qC2 φ2
φ̇L −1 0 1 H(qC , φL ) = + L,
∂φL
" ∂H # 2C 2L

y= 0 1
∂qC the total electromagnetic energy.
∂H
∂φL

Control
Recall the dissipation obstacle

∂F
R(x) (x) = 0.
∂x
For the parallel RLCS circuit it becomes:
∂F
(qC , φL ) = 0 ⇒ F = F (φL ).
∂qC
It’s only possible to shape the energy in the direction of one coordinate (φL ). This is
known as the dissipation obstacle.
L ∗
Physical interpretation: admissible equilibria are of the form qC∗ = CV ∗ , φL = R
V
φ
for any constant V ∗. The power delivered by the source, V ∗ LL ,
is nonzero at any
equilibrium except for the trivial one. Hence, the source has to provide an infinite
amount of energy to keep any nontrivial equilibrium point (Impossible). Notice that pure
mechanical systems are free of this problem: any equilibrium point has velocities equal
to zero.

Control
Overcome the dissipation obstacle using a
State modulated control by interconnection (state modulated source)
IDA-PBC objective
Find a static state-feedback control u(x) = β(x) such that the closed-loop dynamics is
a PH system with interconnection and dissipation of the form
∂Hd
ẋ = (Jd − Rd ) ,
∂x
Hd (x), has a strict local minimum at x ∗ ,

Jd (x, u) = −Jd (x, u)T , the desired interconnection matrix,
Rd (x, u) = Rd (x, u)T ≥ 0, the desired dissipation matrix,

Control
The procedure consists in the matching of the open and (desired) closed-loop vector
fields:
∂H ∂Hd
J(x) − R(x) (x) + g(x)β(x) = Jd (x) − Rd (x) (x)
∂x ∂x
with u = β(x) the state modulated source. Define
Ja = Jd − J, Ja = −Ja>
Ra = Rd − R, Ra = Ra> ≥ 0
Ha = Hd − H
Then the matching condition becomes
∂H ∂Ha
Ja (x) − Ra (x) (x) + g(x)β(x) = J(x) + Ja (x) − R(x) + Ra (x) (x)
∂x ∂x
with design parameters Ja , Ra and Ha .

IDA-PBC: Proposition
Suppose that the following PDE is verified
∂H ∂Ha
g ⊥ (x) Ja (x) − Ra (x) (x) = g ⊥ (x) J(x) + Ja (x) − R(x) + Ra (x)

(x)
∂x ∂x
where g ⊥ (x)g(x) = 0 and Hd (x) is such that
∂Hd ∗ ∂ 2 Hd ∗
(x ) = 0, (x ) > 0,
∂x ∂x 2
then the control

∂H ∂H
u = β(x) = (g > g)−1 g > (Jd − Rd ) d − (J − R)
∂x ∂x
is such that the closed-loop system takes the PH form
∂Hd
ẋ = (Jd − Rd ) ,
∂x
And closed-loop energy
∂Hd> ∂Hd
Ḣd = − Rd < 0, ∀x 6= x ∗ and Ḣd (x ∗ ) = 0.
∂x ∂x
Control
Notice that for systems linear in the control may be interpreted as the optimal solution
to the linear least square problem

∂Hd ∂H
g(x)β(x) = Jd (x) − Rd (x) (x) − J(x) − R(x) (x)
∂x ∂x
| {z }
b
with
β(x) = g † b, subject to the ME (I − g † g)b = 0
where
g † = (g > g)−1 g > : Moore-Penrose pseudo-inverse of g and g † g > projector into rang(g)
(I − g † g) : family of anihilators of g and projector into ker (g)

Remarks
Degrees of freedom in the design

• Jd and Rd are free–up to the constraint of skew–symmetry and positive
semidefiniteness, respectively.
∂H
• Hd may be totally, or partially fixed, provided we can ensure
∂x
d
(x ∗ ) = 0,
∂ 2 Hd
∂x 2
(x ∗ ) ≥ 0 and probably a properness condition.
• there is an additional degree of freedom in g ⊥ (x) which is not uniquely defined by
g(x).
Attention: Requires the solution of a quasilinear PDE
the method of characteristics...

Control
Notice that the method can be equally applied to general input affine non-linear
systems
∂Hd
f (x) + g(x)β(x) = Jd (x) − Rd (x) (x)
∂x
with f (x) a non-linear vector field.
Then the matching condition becomes

∂Hd
g ⊥ (x)f (x) = g ⊥ (x) Jd (x) − Rd (x) (x)
∂x
and the control

∂H
u = β(x) = (g > g)−1 g > (Jd − Rd ) d − f (x)
∂x

Integral Action
IDA–PBC with integral action u = β(x) + v

∂Hd
v̇ = −KI g(x) (x)
∂x
with KI = KI> > 0, preserves stability.
Indeed, define
1 > −1
W (x, v ) = Hd + v KI v
2
The closed-loop system is given by a power preserving interconnection
" ∂Hd #
ẋ J − Rd gKI
= d > > ∂x
∂W
v̇ −KI g 0
∂v
which is again a PHS.

Existing Approaches to Solve the Matching Equation
Non–Parameterized IDA One extreme case

• fix Jd (x), Rd (x) and g ⊥ (x),
• (ME) becomes a PDE for Hd (x),
• among the family of solutions select one that assigns minimum.
Algebraic IDA At the other extreme:
• fix Hd (x),
• (ME) becomes an algebraic equation in Jd (x), Rd (x) and g ⊥ (x),
Parameterized IDA Restrict the desired energy function to a certain class,
• for instance, for mechanical systems
1 > −1 1
Hd (q, p) = p Md (q)p + Vd (q),
2 2
• (ME) becomes a PDE in Md (q) and Vd (q),
• imposes some constraints on Jd (x) and Rd (x).
Application of Poincare’s Lemma (applicable for systems NL in u):
∇H(x) = Fd−1 (x)f (x, u).

Example: parallel RLCS circuit
Consider the following parallel RLC circuit
with x = [qC , φL ], the charge in

the capacitor and the flux in the
inductance
The PH model is
1 " ∂H #
with

q̇C −R 1 ∂qC 0
= ∂H + u qC2 φ2L
φ̇L −1 0 1 H(qC , φL ) = + ,
∂φL
" ∂H # 2C 2L

y= 0 1
∂qC the total electromagnetic energy.
∂H
∂φL

Parallel RLCS circuit: matching equation
Choice of structure matrices and associated added Hamiltonian Ha (x):

1. choose added structure matrices

0 0 0 0
Ja (x) = and Ra (x) = Ra > −R
0 0 0 Ra
2. solve a PDE in Ha (x) using the left annihilator g ⊥ (x) =

1 0 of

0
g (x) = , the matching equation :
1
∂H0 ∂Ha
−g ⊥ (x) (Ja − Ra ) (x) = g ⊥ (x) [(J (x) + Ja (x)) − (R (x) + Ra (x))] (x)
∂x ∂x
becomes:
∂Ha
!
∂qc 1 ∂Ha ∂Ha
− R1

0= 1 ∂Ha =− +
∂φL R ∂qc ∂φL

Parallel RLCS circuit: closed-loop equilibria
The solutions of the matching equation are:
Ha (qc , φL ) = Φ (R qc + φL ) Φ ∈ C ∞ (R)
hence the closed-loop Hamiltonian is
qC2 φ2L
Hd (qC , φL ) = H (qC , φL ) + Φ (R qc + φL ) = + + Φ (R qc + φL )
2C 2L
The equilibrium in closed-loop is given by:
φ∗L qC∗ φ∗L R2C

∂Φ
R − =0 and + 1+ φ∗L = 0
L C L ∂ξ L

Parallel RLCS circuit: two possible H
ξ2
1. Case 1: Φ (ξ) = k 2
then:
L + R2C

φ∗ ∈ R and k =− <0
L2
and Φ (ξ) is concave !
ξ4
2. Case 2: Φ (ξ) = k 4
then:
1
φ∗ = ± q and k =∈ R∗−
(−k) L + R 2 C
and Φ (ξ) is again concave !

Parallel RLCS circuit: local positivity of Hd
Consider the Hessian of Hd at the equilibrium:

 2 2 
1
+ R 2 ∂∂ξΦ2 Rqc∗ + φ∗2 R ∂∂ξΦ2 Rqc∗ + φ∗2

∂ 2 Hd  C L L
= 2
∂2Φ

∂qc , φ2L R ∂ Φ2 Rqc∗ + φ∗2 1
Rqc∗ + φ∗2

∂ξ L L
+ ∂ξ2 L
is definite positive iff:

2 2
1
+ R 2 ∂∂ξΦ2 Rqc∗ + φ2L > 0 or: 1L + ∂∂ξΦ2 Rqc∗ + φ∗2

1. either: C L >0
∂ 2 Hd 2

1
1 + L + R 2 C ∂∂ξΦ2 Rqc∗ + φ∗2

2. and det 2 > 0 i.e. : LC L >0
∂qc ,φL
∂2Φ 1
Rqc∗ + φ2L > −

which reduces to: ∂ξ2 (R 2 C+L)

Parallel RLCS circuit: two possible Φ
∂2Φ 1
Rqc∗ + φ2L > −

Check for the two examples the condition : ∂ξ2 (R 2 C+L)
ξ2
1. Case 1: Φ (ξ) = k 2
the condition reduces to:
2
L + R2C < L2
which is wrong !
ξ4
2. Case 2: Φ (ξ) = k 4
then:
1
(−k) <
φ∗L 2
L + R2C L
which leads to a solution !

Parallel RLCS circuit: IDA-PBC control
The control law is given by :

t −1 t
β (x) = ng (x) g (x) g (x) o
∂Ha ∂H
[(J (x) + Ja (x)) − (R (x) + Ra (x))] ∂x
(x) + (Ja − Ra ) ∂x
(x)
which becomes :
" !
∂Φ
− R1 R (R qC , +φL )

1 ∂ξ
β (qC , φL ) = 1 0 1 ∂Φ
−1 −Ra ∂ξ
(R qC , +φL )
qC
!#
0 0 C
− φL
0 Ra
L
or :
∂Φ φL
β (qC , φL ) = (R − Ra ) ∂ξ
(R qC , +φL ) − Ra L

The levitating iron ball in a magnetic field
The mathematical model
φ̇1 = −Ri + u
ẏ = v
mv̇ = −Fm + mg
with u input voltage, R electric resistance,

mthe mass and g the gravitational constant,
∂Wc
Fm = − (i, y) electro-mechanical coupling
∂y
1
Wc = L(y)i 2 non-linear inductance
2
and the non-linear inductance
k
L(y) = L∞ + ,
(a + y)
with L∞ , a, k > 0 and φ = L(y)i.

The levitating iron ball: the PH model
The port-Hamiltonian model

x1 = φ total magnetic flux
set as state variables: x2 = y displacement of the ball
x3 = mv = p kinetic momentum
 
     
 0 0 0 R 0 0  1
dx   ∂H
=  0 0 1 −  0 0  ∂x (x) + 0 u
0  
dt 
 0 −1 0 0 0 0  0
| {z } | {z } |{z}
J R g
∂H ∂H
y = 1 0 0 (x) = =i
∂x ∂x1
1 1
with total energy: H(x) = x 2
2 L(x2 ) 1
+ x 2
2m 3
− mgx2
 x1 
L(x2 ) current through the coil, i
∂H0  x3 
(x) =  m
 velocity of the ball
∂x 
x1
2 
electro-motive+gravity force
− 12 dx
dL
(x2 ) L(x )
−mg
2 2

The levitating ball: coupling through energy
The coupling between the mechanical and magnetic domain does not occur through
the structure matrices
 
−R 0 0 magnetic domain
J −R =  0 0 1 mechanical potential domain
0 −1 0 mechanical kinetic domain
The structure matrix is block-diagonal.
The coupling occurs through the Hamiltonian which is not

separated
1 1
H0 (x) = x1 2 + x3 2 − mgx2
2 L (x2 ) 2m

The levitating ball: open-loop equilibria
The equilibria are given by (x ∗ , u ∗ ) such that:

x1∗
 
R + u∗
L(x2∗ )
x3∗
 
∂H ∗
(x ) + g u ∗ = 
 
(J − R) =0
∂x  m 2 
 x1∗ 
− 21 dx
dL
x2∗

−mg
2 L(x2∗ )
∂H
and may be parametrized by the current: y ∗ = ∂x (x ∗ ) = i ∗ :
1
L∞ y ∗ +q 2mg
 ∗   p 
x1
∗
 x  =  −a + k y ∗  and u ∗ = R y ∗
 
2 2mg
x3∗ 0
they are unstable (see linearised system).

The levitating ball: matching equation
Choice of structure matrices and associated added Hamiltonian Ha (x):

1. choose added structure matrices:
 
0 0 α
Ja (x) =  0 0 0  and Ra (x) = 03
−α 0 0
What’s the physical interpretation?

0 1 0
2. Use the left annihilator g ⊥ (x) = , establish the PDE in Ha :
0 0 1
∂H0
   
∂Ha
∂x1 ∂x1
0 0 0  ∂H0  0 0 1  ∂Ha 
−  = 
∂x2

−α 0 0  ∂x2  −α −1 0 
∂Ha

∂H0
∂x3 ∂x3

The levitating ball: added Hamiltonian
From : ∂H
∂x3
a
= 0 the added potential is a function Ha (x1 , x2 ) which does not depend on
the velocity and satisfies :
∂H0 ∂Ha ∂Ha
α = −α −
∂x1 ∂x1 ∂x2
with solution:
Z x1 χ x
Ha (x1 , x2 ) = − dχ + Φ x2 − 1
(χ−x1 ) α
0 L x2 − α
For instance, if L (x2 ) = (x k+a) then

3 2
1 x1 2 (x + a) + Φ x − x1
Ha (x1 , x2 ) = 2k 3α
− x 1 2 2 α

The levitating ball: IDA-PBC control
The control law is given by :

 
−R 0 α
1  ∂H
a
β (x) = 1 1 0 0  0 0 ∂x
(x)
| {z
−1
} −α −1 0
=[ g t g ] g t
  
0 0 α
∂H
+ 0 0 0  ∂x0 (x)
−α 0 0
or :
∂Ha ∂Ha ∂H0
β (x) = −R (x) + α (x) + α (x)
∂x1 ∂x3 ∂x3

The levitating iron ball: the closed-loop system
 
   
 0 0 α R 0 0 
dx   ∂Hd
 0
=  0 1 −  0 0 0 (x)
dt  −α −1 0 0 0 0  ∂x

| {z } | {z }
J R

Remarks
Hamiltonian methods for the control of physical systems:

1. use the interconnection of Port Hamiltonian systems and Casimir functions
2. assign closed-loop Hamiltonian function and structure matrices.
A control synthesis based on insight of desired physical behaviour in closed-loop:
1. design directly interconnection of the system with environment and indirectly the
controller
2. design the closed-loop port Hamiltonian behaviour and deduce the controller
Open questions:
1. choice of desired behaviour in closed-loop and relation with performance and
robustness
2. parametrization of the matching equations and solution

Concluding remarks
• Energy based modelling: based on the universal concept of energy transfer.

• Provides physical interpretation to the models and the solutions.
• Passivity is naturally encountered when working with problems arising from
physical applications.
• Port-Hamiltonian control systems defines a class of non-linear passive systems
which encompasses a large class of physical applications.
• A modelling and control approach which is transversal to different (or combination
of) physical domains.

Stability

Uploaded by

Copyright:

Available Formats

Stability

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Stability

Uploaded by

Copyright:

Available Formats

Introduction to Control of port-Hamiltonian systems

Héctor Ramı́rez, Yann Le Gorrec and Yongxin Wu

AS2M, FEMTO-ST UMR CNRS 6174,

22-24 March, 2017

Doctoral course UFC-ST 1 / 103

Doctoral course UFC-ST 2 / 103

4. Interconnection and Damping Assignment – Passivity Based Control

5. Final remarks and future work

Doctoral course UFC-ST 3 / 103

We are considering the following class of state space model

with t ∈ R and x ∈ Rn . Furthermore f : Rn × Rm → Rn . It is assumed that f (x(t))

We want to analyse the dynamics of the system

• Do the solutions of (1) remain bounded in time?

Doctoral course UFC-ST 4 / 103

We are considering the following class of state space model

with t ∈ R and x ∈ Rn . Furthermore f : Rn × Rm → Rn . It is assumed that f (x(t))

Without loss of generality we will assume that x ∗ = 0.

Doctoral course UFC-ST 5 / 103

An equilibrium position x = 0 of system (1) is

lim kx(t, x0 )k → 0 (3)

4. Uniformly Asymptotically Stable if it is uniformly stable, ∆ does not depend on t0 ,

Doctoral course UFC-ST 6 / 103

An equilibrium position x = 0 of system (1) is

kx(t, x0)k < Mkx0 ke−α(t−t0 ) (4)

for all t ≥ t0 ≥ 0, and

Doctoral course UFC-ST 7 / 103

Figures taken from:

Doctoral course UFC-ST 8 / 103

How do we analyse stability?

Lyapunov stability theory

• Lyapunov’s direct method (second method) → non-linear systems

Doctoral course UFC-ST 9 / 103

Let B be a ball of size  around the origin, B = x ∈ Rn : x < kk.

Positive definite function

A positive definite function is like an energy function.

Doctoral course UFC-ST 10 / 103

Theorem: Lyapunov stability Let V (x) be a non-negative function with continuous

Theorem: Asymptotic stability Let V (x) be a non-negative function with continuous

Doctoral course UFC-ST 11 / 103

Figure: taken from: http://www.math24.net/

Doctoral course UFC-ST 12 / 103

So why do we need the control then?

Doctoral course UFC-ST 13 / 103

• What if we want to increase the rate of convergence?: damping injection,

Doctoral course UFC-ST 14 / 103

Consider the system

ẋ(t) = f (x(t), u(t)), y(t) = h(x(t), u(t)), (5)

with t ∈ R, x ∈ Rn , u ∈ Rm and y ∈ Rp . Furthermore f : Rn × Rm → Rn and

Doctoral course UFC-ST 15 / 103

A particular case of dissipative systems are passive systems:

Then the system is passive.

Doctoral course UFC-ST 16 / 103

Strictly passive systems

The function S(x) is called the the dissipation rate.

Doctoral course UFC-ST 17 / 103

Why are we interested in passive systems?

Doctoral course UFC-ST 18 / 103

Recall the passivity inequality

if the storage function V is strictly increasing and x = 0 is an isolated global minimum,

For many physical systems: Energy = storage function = Lyapunov function

Let B be a ball of size around the origin, B = x ∈ Rn : x < kk.

dH(x(t)) ∂H > dx q p q p p p 2