SlideShare a Scribd company logo
Continuous Finite-Time Stabilization of the
Translational and Rotational Double Integrator
Olalekan P. Ogunmolu
April 30, 2015
1
Introduction
Feedback linearization often generates closed-loop Lipschitzian dynamics
Convergence in such systems is often exponential, carrying the burden of infinite settling time
Undesirable in time-critical applications such as minimum-energy control, conservation of momentum et cetera
It is therefore imperative to design continuous finite-time stabilizing controllers
Classical optimal control provides examples of systems that converge to the equilibrium in finite time such as the
double integrator.
2
The case for continuous finite-time stabilizing (FTS) con-
trollers?
Optimal synthesis methods such as the bang-bang time-optimal feedback control of the double integrator
has discontinuous dynamics which leads to chattering or excitation of HF dynamics in flexible structures
applications [Fuller et. al., ’66]
The double integrator is open-loop controllable such that its states can be driven to the origin in finite time
The minimum-energy control is one such method which drives the state of the system ¨x = u from an
initial condition x(0) = x0, ˙x(0) = y0 to the origin in a given time tf [Athans, ’66]
But such open-loop methods invite insensitivity to uncertainties and have poor disturbance rejection
Non-Lipschitzian Dynamics of a Continuous FTS Feedback Controller
1
Overview of Problem
The design of FTS continuous time-invariant feedback controllers involve non-Lipschitzian closed-loop dynamics
Such controllers will exhibit non-unique solutions in backward time, i.e., better robustness and good disturbance
rejection
Such non-unique (revert time) solutions would violate uniqueness conditions for Lipschitz differential equations
2
Statement of Problem
Consider a rigid body rotating under the action of a mechanical torque about a fixed axis
Its equations of motion resemble those of a double integrator. States differ by 2nπ (where
n = 0, 1, 2, . . .) in angular modes which correspond to the same physical configuration of the body.
State space for this system is S1
× R rather than R2
[Andronov et. al]
Developing stabilizing controls for the double integrator on R2
(translational double integrator) will lead to
unwinding since the configuration space is actually R
This makes an interesting problem when designing feedback controllers for the rotational double
integrator with anti-wind-up compensation
Discontinuous feedback controllers are practically infeasible due to the chattering they introduce
because of plant uncertainties
They could also excite high-frequency dynamics when used in controlling lightly damped structures
[Baruh et. al.]
1
Finite-Time Stabilization: A Definition
For the System of differential equations,
˙y(t) = f(y(t) (1)
where f : D → Rn
is continuous on an open neighborhood D ⊆ Rn
of the origin and f(0) = 0. A continuously differentiable
function y : I → D is said to be a solution of 1 on the interval I ⊂ R if y satisfies (1) for all t ∈ I
We assume (1) possesses unique solutions in forward time except possibly at the origin for all initial conditions
Uniqueness in forward time and the continuity of f ensure that solutions are continuous functions of initial conditions
even when f is no longer Lipschitz continuous [Hartman et. al., ’82, Th. 2.1, p. 94]
2
Definition
The origin is finite-time stable if there exists an open neighborhood N ⊆ D of the origin and a settling time
function T : N  0 → (0, ∞), such that we have the following:
1. Finite-time convergence: For every x ∈ N  {0}, ρt (x) is defined for
t ∈ 0, T(x) , ρt (x) ∈ N  {0}, for t ∈ 0, T(x) , and limt→T(x)ρt (x) = 0
2. Lyapunov stability: For every open set U such that 0 ∈ U ⊆ N , there exists an open set Uδ such
that 0 ∈ Uδ ⊆ N and for every x ∈ Uδ  {0}, ρt (x) ∈ U• for t ∈ 0, T(x) .
When D = N = Rn
, we have global finite-time convergence.
Theorem: For a continuously differentiable function V : D → R, such that k > 0, α ∈ (0, 1) , where α and k ∈ R if there
exists a neighborhood of the origin U ⊂ D such that V is positive definite, ˙V is negative definite and ˙V + kVα
is negative
semi-definite on U, where ˙V(x) =
∂V
∂x
(x)f(x). Then, the origin of (1) is finite-time stable. Also, the settling time, T(x) is
defined as T(x) =
1
k(1 − α)
V(x)1−α
1
Continuous Finite-Time Stabilizing Controllers
We want to find a continuous feedback law, u = ψ(x, y) such that the double integrator defined as, ˙x = y, ˙y = u is
finite-time stabilized.
2
Proposition 1
The origin of the double integrator is globally finite-time stable [Bhat et. al., ’98, §III] under the feedback control
law u where
ψ(x, y) = −sign (y)|y|
α
− sign φα (x, y) φα (x, y)
α
2 − α (2)
where φα(x, y) x +
1
2 − α
sign(y)|y|
α
2 − α
See Appendix A. for proof.
3
Remarks
The vector field obtained by using the feedback control law u is locally Lipschitz everywhere except
the x-axis (denoted Γ ), and the zero-level set S = {(x, y) : φα(x, y) = 0} of the function φα. The
closed-loop vector field fα is transversal to Γ at every point in Γ {0, 0}
Every initial condition in Γ {0, 0} has a unique solution in forward time
The set S is positively invariant for the closed-loop system
On the set S the closed-loop system is
˙x = −sign (x) (2 − α)|x|
1
2−α (3)
˙y = −sign (y)|y|
α
(4)
The resulting closed loop system (4) is locally Lipschitz everywhere except the origin and therefore
possesses unique solutions in forward time for initial conditions in S  {0, 0}.
1
Example 1: Implementation of the proposed controller
By choosing α = 2
3
in (2), we have the phase portrait shown in Figure 6 for the resulting feedback law
ψ(x, y) = −y
2
3 − x +
3
4
y
4
3
1
2
(5)
We see that all trajectories converge to the set S = {(x, y) : x + 3
4
y
4
3 = 0} in finite-time. The term −y
2
3 in (5) makes the
set S positively invariant while the other term − x + 3
4
y
4
3
1
2
drives the states to S in finite-time. Therefore, (2) represents
an example of a terminal sliding mode control without using discontinuous or high gain feedback.
1
Bounded Continuous Finite-Time Controllers
In the previous section, the designed feedback controller is unbounded, meaning the controller will lead to the so-called
”unwinding” phenomenon.
In a spacecraft application, for example, such unwinding can lead to the mismanagement of fuel and and
momentum-consuming devices
In order to finite-time stabilize the controller, we saturate the elements of the controller given in (2)
It follows that for a positive number ε,
satε (y) = y, |y| < ε (6)
= ε sign (y) , |y| ≥ ε
such that satε (y) ≤ ε for all y ∈ R (7)
The relation in (7) ensures that the controller does not operate in the nonlinear region by bounding its operating range within
the defined perimeter of operation (6)
1
Proposition 2
The origin of the double integrator system is a globally stable equilibrium under the bounded feedback control law u =
ψ (x, y) with
ψsat (x, y) = −satε {sign (y)|y|
α
} − {satε sign φα (x, y) φα (x, y)
α
2 − α } (8)
for every α ∈ (0, 1) and ε > 0, where φα(x, y) x +
1
2 − α
sign(y)|y|
α
2 − α
See Proof in [Bhat et. al., ’98, §IV, p. 680].
1
Example 2: Bounded continuous finite time controller
For the double integrator under the feedback control law
ψsat (x, ) = −sat1(y
1
3 ) − sat1{(x +
3
5
y
5
3 )}
1
5 (9)
obtained from (2) with α = 1
3
and ε = 1. Again, all trajectories converge to the set S = {(x, y) : x + 3
5
y
5
3 = 0}. But in
some phase plane regions, ψsat (x, ) = 0.
1
The Rotational Double Integrator
Let us denote the motion of a rigid body rotating about a fixed axis with unit moment of inertia as
¨θ(t) = u(t) (10)
where θ is the angular displacement from some reference and u is the applied control. We can rewrite the equation as a
first-order equation with ˙x = θ and ˙y = u.
Suppose we require the angular position to be finite-time stable, then the feedback law given in (2) can only
finite-time stabilize the origin such that if applied to the rotational double integrator, it leads to the unwinding
phenomenon.
Feedback controllers designed for the translational double integrator does not suffice for the rotational double
integrator
How to avoid the unwinding phenomenon?
Modify (2) such that it is periodic in x with period 2π, i.e.,
ψrot (x, y) = −sign (y)|y|
α
− sign (sin(φα(x, y))) sin(φα(x, y))
α
2−α (11)
φα is same as defined in Proposition 1. For α = 1
3
with u = ψrot the resulting phase portrait is shown below
1
Remarks from Phase Portrait of Rotational Double Integrator
The closed loop system has equilibrium points at sn = (2nπ, 0), un = ((2n + 1)π, 0), n = · · · , −1, 0, 1, · · ·
The equilibrium points sn are locallu finite-time stable in forward time, while the points un are finite-time saddles
The domain of attraction of the equilibrium point s is Dn = {(x, y) : (2n − 1)π < φα(x, y) < (2n + 1)π}.
The shaded region of the plot shows a portion of D0. The sets Un−1 and Un represent the stable manifolds of the
equilibrium points un−1 and un respectively where Un = {(x, y) : φα(x, y) = (2n + 1)π} and
n = · · · , −1, 0, 1, · · · ,
All trajectories starting in the set Dn converge to the set Sn = {(x, y) : φα(x, y) = 2nπ} in finite, forward time
and to the set Un−1 ∪ Un in finite, reverse time
The sets Sn are positively invariant while the sets Un are negatively invariant
2
The solutions in the figure have no uniqueness to initial conditions lying in any of the sets
Un, n = · · · , −1, 0, 1, · · ·.
All solutions initialized in un are equivalent to the rigid body resting in an unstable configuration and then
starting to move spontaneously clockwise or counterclockwise
Departure from the unstable equilibrium is a unique feature to non-Lipschitzian systems as Lipschitzian
systems do not possess solutions that depart from equilibrium
The desired final configuration is however not globally finite-time stable due to the presence of the
unstable equilibrium configuration at θ = π. These are saddle points un, n = · · · , −1, 0, 1, · · ·
This is a basic drawback to every continuous feedback controller that stabilizes the rotational double
integrator without generating the unwinding effect.
The desired final configuration in the phase plane corresponds to multiple equilibria in the phase plane
meaning every controller that stabilizes the desired configuration stabilizes each equilibria
But stability, continuous dependence on initial conditions and solutions’ uniqueness imply that the
domain of attraction of any two equilibrium points in the plane are non-empty, open and disjoint.
1
We cannot write R2
as the union of a collection of disjoint sets.
Thus, there are initial conditions in the plane that do not converge to the equilibria of the desired final configuration.
With respect to (11), these initial conditions are the stable manifolds of the unstable configuration.
The designed controller is practically globally stable as its non-Lipschitzian property increases the sensitivity of the
unstable configuration to perturbations
1
Appendix A: Proof of Proposition 1
If we denote φα(x, y) by φα and fix α ∈ (0, 1), we could choose the C2
Lyapunov function candidate,
V (x, y) =
2 − α
3 − α
|φα|
3 − α
2 − α + syφα +
r
3 − α
|y|
3−α
(12)
where r and s are positive numbers. Along the closed loop trajectories,
˙V (x, y) = sφα ˙y + r|y|
2−α
˙y + sy ˙φα +|φα|
1
2 − α ˙φα = sφα



−sign (y)|y|
α
− sign (φα)|φα|
α
2 − α



 + r|y|
2−α




(13)
+|φα|
1
2 − α ˙φα
But, ˙φα = ˙x + ˙y sign (y) |y|
1−α
(14)
from (2), it therefore follows that,
1
Continuation of Proof of Proposition 1
˙φα = −sign (y) sign (φα)|y|1−α
|φα|
α
2−α (15)
Putting (14) into (13), and noting that
sy ˙φα = −s sign (yφα)|y|
1−α
|φα|
1 + α
2 − α (16)
and ˙φα|φα|
1
2−α = −sign (y) sign (φα) |y|
1−α
|φα|
1 + α
2 − α (17)
we find that,
(18)˙V (x, y) = −ry
2
− s|φα|
2
2−α −|y|
1−α
|φα|
1+α
2−α − sφαsign (y)|y|
α
− (r + s)sign(yφα)|y|
2−α
|φα|
α
2−α
1
Remarks
The obtained Lyapunov derivative in the foregoing is continuous everywhere since α ∈ (0, 1). For k > 0 and (x, y) ∈ R2
the following holds,
Introduce x = k2−α
, y = ky such that
1
Remarks
(19)
φα(k
2−α
x, ky) = k
2−α
x −
1
2 − α
sign(ky)|ky|
2−α
= k
2−α
φα(x, y))
and
(20)V(k
2−α
x, ky) =
2 − α
3 − α
φα(k
2−α
x, ky)
3 − α
2 − α + skyφα(k
2−α
x, ky) +
r
3 − α
|ky|
3−α
such that (21)V(k
2−α
x, ky) = k
3−α
V(x, y)
Following a similar logic as in (20), we find that
(22)˙V(k
2−α
x, ky) = k
2 ˙V(x, y)
The results of the previous section imply that for r > 1 and s < 1, both V and ˙V are positive on the set O =
{(x, y) : max(x,y)=(0,0)|φα|
1
2−α ,|y| = 1} which is a closed curve encircling the origin.
For every (x, y) ∈ R2
 {0, 0} there exists k > 0 such that k2−α
x, ky ∈ O, the homogeneity properties of (20) and (22)
imply a positive definite V and negative semi-definite ˙V.
From (20), V is radially unbounded so that the set V = {(x, y) : V(x, y) = 1} is compact. Therefore ˙V achieves its
maximum on the compact set V. If we define c = −max{(x,y)∈V}
˙V(x, y), then ˙V(x, y) ≤ −c{V(x, y)}
2
3−α for all
(x, y) ∈ R2
[Bhat et. al. ’96]
The homogeneity of (20) and (22) ensures ˙V(x, y) ≤ −cV(x, y)
2
3−α for all (x, y) ∈ R2
[Bhat et. al. ’96]. Since α ∈ (0, 1)
is ≡ 2
3−α
∈ (0, 1), we can conclude finite time stability
1
References
A. A. Andronov, A. A. Vitt, and S. E. Khaikin. Theory of Oscillators. Oxford, UK: Pergamon, 1966.
M. Athans and P.L. Falb. Optimal Control: An Introduction to the Theory and Its Applications. McGraw-Hill, New York,
1966.
H. Baruh and S.S.K. Tadikonda. Gibbs phenomenon in structural control. J. Guidance, Contr., and Dynamics. vol.,
no. 1, pp. 51-58, 1991.
S.P. Bhat and D.S. Bernstein. Continuous Finite-Time Stabilization of the Translational and Rotational Double Integra-
tors. IEEE Transactions on Automatic Control. vol. 43., no. 5, 1998
S.P. Bhat and D.S. Bernstein. Continuous, Bounded, Finite-Time Stabilization of the Translational and Rotational
Double Integrators. in Proc. Amer, Contr. Conf., Seattle, WA, June 1996, pp. 1831 - 1832
A. Fuller. Optimization of some nonlinear control systems by means of Bellman’s equation and dimensional analysis.
Int. J. Contr., vol. 3, no. 4, pp. 359–394, 1966.
P. Hartman. Ordinary Differential Equations. 2nd edition, Boston, MA: Birkhauser, 1982.

More Related Content

What's hot

Equation of motion of a variable mass system2
Equation of motion of a variable mass system2Equation of motion of a variable mass system2
Equation of motion of a variable mass system2
Solo Hermelin
 
Keplerian trajectories
Keplerian trajectoriesKeplerian trajectories
Keplerian trajectories
Solo Hermelin
 
4 stochastic processes
4 stochastic processes4 stochastic processes
4 stochastic processes
Solo Hermelin
 
Notes mech v
Notes mech vNotes mech v
Notes mech vRung Heo
 
Lecture 5 castigliono's theorem
Lecture 5 castigliono's theoremLecture 5 castigliono's theorem
Lecture 5 castigliono's theoremDeepak Agarwal
 
Response spectra
Response spectraResponse spectra
Response spectra
321nilesh
 
Convolution
ConvolutionConvolution
Convolution
muzuf
 
Slide Mode Control (S.M.C.)
Slide Mode Control (S.M.C.)Slide Mode Control (S.M.C.)
Slide Mode Control (S.M.C.)
Solo Hermelin
 
Single degree of freedom system free vibration part -i and ii
Single degree of freedom system  free vibration part -i and iiSingle degree of freedom system  free vibration part -i and ii
Single degree of freedom system free vibration part -i and ii
Sachin Patil
 
One particle to_onepartlce_scattering_18052020
One particle to_onepartlce_scattering_18052020One particle to_onepartlce_scattering_18052020
One particle to_onepartlce_scattering_18052020
foxtrot jp R
 
5 earth gravitation
5 earth gravitation5 earth gravitation
5 earth gravitation
Solo Hermelin
 
One particle to_onepartlce_scatteringsqrdcpy1
One particle to_onepartlce_scatteringsqrdcpy1One particle to_onepartlce_scatteringsqrdcpy1
One particle to_onepartlce_scatteringsqrdcpy1
foxtrot jp R
 
Important Notes - JEE - Physics - Simple Harmonic Motion
Important Notes - JEE - Physics - Simple Harmonic MotionImportant Notes - JEE - Physics - Simple Harmonic Motion
Important Notes - JEE - Physics - Simple Harmonic Motion
Ednexa
 
Operators n dirac in qm
Operators n dirac in qmOperators n dirac in qm
Operators n dirac in qmAnda Tywabi
 
One particle to_onepartlce_scattering_5302020_pdfcpy
One particle to_onepartlce_scattering_5302020_pdfcpyOne particle to_onepartlce_scattering_5302020_pdfcpy
One particle to_onepartlce_scattering_5302020_pdfcpy
foxtrot jp R
 
One particle to_onepartlce_scattering_12082020_fordisplay
One particle to_onepartlce_scattering_12082020_fordisplayOne particle to_onepartlce_scattering_12082020_fordisplay
One particle to_onepartlce_scattering_12082020_fordisplay
foxtrot jp R
 

What's hot (20)

Equation of motion of a variable mass system2
Equation of motion of a variable mass system2Equation of motion of a variable mass system2
Equation of motion of a variable mass system2
 
Keplerian trajectories
Keplerian trajectoriesKeplerian trajectories
Keplerian trajectories
 
4 stochastic processes
4 stochastic processes4 stochastic processes
4 stochastic processes
 
Notes mech v
Notes mech vNotes mech v
Notes mech v
 
M1l3
M1l3M1l3
M1l3
 
Lecture 5 castigliono's theorem
Lecture 5 castigliono's theoremLecture 5 castigliono's theorem
Lecture 5 castigliono's theorem
 
Z transform
Z transformZ transform
Z transform
 
Response spectra
Response spectraResponse spectra
Response spectra
 
Bazzucchi-Campolmi-Zatti
Bazzucchi-Campolmi-ZattiBazzucchi-Campolmi-Zatti
Bazzucchi-Campolmi-Zatti
 
Convolution
ConvolutionConvolution
Convolution
 
Slide Mode Control (S.M.C.)
Slide Mode Control (S.M.C.)Slide Mode Control (S.M.C.)
Slide Mode Control (S.M.C.)
 
Single degree of freedom system free vibration part -i and ii
Single degree of freedom system  free vibration part -i and iiSingle degree of freedom system  free vibration part -i and ii
Single degree of freedom system free vibration part -i and ii
 
One particle to_onepartlce_scattering_18052020
One particle to_onepartlce_scattering_18052020One particle to_onepartlce_scattering_18052020
One particle to_onepartlce_scattering_18052020
 
5 earth gravitation
5 earth gravitation5 earth gravitation
5 earth gravitation
 
One particle to_onepartlce_scatteringsqrdcpy1
One particle to_onepartlce_scatteringsqrdcpy1One particle to_onepartlce_scatteringsqrdcpy1
One particle to_onepartlce_scatteringsqrdcpy1
 
Important Notes - JEE - Physics - Simple Harmonic Motion
Important Notes - JEE - Physics - Simple Harmonic MotionImportant Notes - JEE - Physics - Simple Harmonic Motion
Important Notes - JEE - Physics - Simple Harmonic Motion
 
Operators n dirac in qm
Operators n dirac in qmOperators n dirac in qm
Operators n dirac in qm
 
One particle to_onepartlce_scattering_5302020_pdfcpy
One particle to_onepartlce_scattering_5302020_pdfcpyOne particle to_onepartlce_scattering_5302020_pdfcpy
One particle to_onepartlce_scattering_5302020_pdfcpy
 
One particle to_onepartlce_scattering_12082020_fordisplay
One particle to_onepartlce_scattering_12082020_fordisplayOne particle to_onepartlce_scattering_12082020_fordisplay
One particle to_onepartlce_scattering_12082020_fordisplay
 
SDEE: Lecture 6
SDEE: Lecture 6SDEE: Lecture 6
SDEE: Lecture 6
 

Viewers also liked

Andrew goodwin
Andrew goodwinAndrew goodwin
Andrew goodwin
kaan88
 
JKempnich - FINC 5380 - Final Research Paper
JKempnich - FINC 5380 - Final Research PaperJKempnich - FINC 5380 - Final Research Paper
JKempnich - FINC 5380 - Final Research PaperJacob Kempnich
 
Bluemix cfmeetup
Bluemix cfmeetupBluemix cfmeetup
Bluemix cfmeetup
Lawrence Crowther
 
Joseph kahn
Joseph kahnJoseph kahn
Joseph kahn
kaan88
 
Coding in the open CF - Cloud Foundry on Azure
Coding in the open CF - Cloud Foundry on AzureCoding in the open CF - Cloud Foundry on Azure
Coding in the open CF - Cloud Foundry on Azure
Lawrence Crowther
 
Settings media
Settings media Settings media
Settings media
kaan88
 
Sydney cloud foundry meetup - Service Brokers
Sydney cloud foundry meetup - Service  BrokersSydney cloud foundry meetup - Service  Brokers
Sydney cloud foundry meetup - Service Brokers
Lawrence Crowther
 
Towards fine-precision automated immobilization in maskless radiosurgery
Towards fine-precision automated immobilization in maskless radiosurgeryTowards fine-precision automated immobilization in maskless radiosurgery
Towards fine-precision automated immobilization in maskless radiosurgery
Olalekan Ogunmolu
 
Pivotal CF on AWS
Pivotal CF on AWSPivotal CF on AWS
Pivotal CF on AWS
Lawrence Crowther
 
Linkedin "link to professionals"
Linkedin "link to professionals"Linkedin "link to professionals"
Linkedin "link to professionals"
Medhat Kilany
 
PhD Qualifying Exam Slides
PhD Qualifying Exam SlidesPhD Qualifying Exam Slides
PhD Qualifying Exam Slides
Olalekan Ogunmolu
 
Laporan fieldtrip geologi dasar
Laporan fieldtrip geologi dasarLaporan fieldtrip geologi dasar
Laporan fieldtrip geologi dasar
Rima Rosaliana
 

Viewers also liked (13)

Andrew goodwin
Andrew goodwinAndrew goodwin
Andrew goodwin
 
JKempnich - FINC 5380 - Final Research Paper
JKempnich - FINC 5380 - Final Research PaperJKempnich - FINC 5380 - Final Research Paper
JKempnich - FINC 5380 - Final Research Paper
 
Jessica´s vacations
Jessica´s vacationsJessica´s vacations
Jessica´s vacations
 
Bluemix cfmeetup
Bluemix cfmeetupBluemix cfmeetup
Bluemix cfmeetup
 
Joseph kahn
Joseph kahnJoseph kahn
Joseph kahn
 
Coding in the open CF - Cloud Foundry on Azure
Coding in the open CF - Cloud Foundry on AzureCoding in the open CF - Cloud Foundry on Azure
Coding in the open CF - Cloud Foundry on Azure
 
Settings media
Settings media Settings media
Settings media
 
Sydney cloud foundry meetup - Service Brokers
Sydney cloud foundry meetup - Service  BrokersSydney cloud foundry meetup - Service  Brokers
Sydney cloud foundry meetup - Service Brokers
 
Towards fine-precision automated immobilization in maskless radiosurgery
Towards fine-precision automated immobilization in maskless radiosurgeryTowards fine-precision automated immobilization in maskless radiosurgery
Towards fine-precision automated immobilization in maskless radiosurgery
 
Pivotal CF on AWS
Pivotal CF on AWSPivotal CF on AWS
Pivotal CF on AWS
 
Linkedin "link to professionals"
Linkedin "link to professionals"Linkedin "link to professionals"
Linkedin "link to professionals"
 
PhD Qualifying Exam Slides
PhD Qualifying Exam SlidesPhD Qualifying Exam Slides
PhD Qualifying Exam Slides
 
Laporan fieldtrip geologi dasar
Laporan fieldtrip geologi dasarLaporan fieldtrip geologi dasar
Laporan fieldtrip geologi dasar
 

Similar to Projekt

Z Transform And Inverse Z Transform - Signal And Systems
Z Transform And Inverse Z Transform - Signal And SystemsZ Transform And Inverse Z Transform - Signal And Systems
Z Transform And Inverse Z Transform - Signal And Systems
Mr. RahüL YøGi
 
The feedback-control-for-distributed-systems
The feedback-control-for-distributed-systemsThe feedback-control-for-distributed-systems
The feedback-control-for-distributed-systemsCemal Ardil
 
Reachability Analysis Control of Non-Linear Dynamical Systems
Reachability Analysis Control of Non-Linear Dynamical SystemsReachability Analysis Control of Non-Linear Dynamical Systems
Reachability Analysis Control of Non-Linear Dynamical SystemsM Reza Rahmati
 
Reachability Analysis "Control Of Dynamical Non-Linear Systems"
Reachability Analysis "Control Of Dynamical Non-Linear Systems" Reachability Analysis "Control Of Dynamical Non-Linear Systems"
Reachability Analysis "Control Of Dynamical Non-Linear Systems" M Reza Rahmati
 
APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...
APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...
APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...
ijfls
 
APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...
APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...
APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...
Wireilla
 
Ph 101-9 QUANTUM MACHANICS
Ph 101-9 QUANTUM MACHANICSPh 101-9 QUANTUM MACHANICS
Ph 101-9 QUANTUM MACHANICS
Chandan Singh
 
Fault tolerant process control
Fault tolerant process controlFault tolerant process control
Fault tolerant process controlSpringer
 
Notes nyquist plot and stability criteria
Notes nyquist plot and stability criteriaNotes nyquist plot and stability criteria
Notes nyquist plot and stability criteria
Aleksandar Micic
 
DSP_FOEHU - MATLAB 02 - The Discrete-time Fourier Analysis
DSP_FOEHU - MATLAB 02 - The Discrete-time Fourier AnalysisDSP_FOEHU - MATLAB 02 - The Discrete-time Fourier Analysis
DSP_FOEHU - MATLAB 02 - The Discrete-time Fourier Analysis
Amr E. Mohamed
 
Adaptive dynamic programming algorithm for uncertain nonlinear switched systems
Adaptive dynamic programming algorithm for uncertain nonlinear switched systemsAdaptive dynamic programming algorithm for uncertain nonlinear switched systems
Adaptive dynamic programming algorithm for uncertain nonlinear switched systems
International Journal of Power Electronics and Drive Systems
 
Controller design of inverted pendulum using pole placement and lqr
Controller design of inverted pendulum using pole placement and lqrController design of inverted pendulum using pole placement and lqr
Controller design of inverted pendulum using pole placement and lqr
eSAT Publishing House
 
Controller design of inverted pendulum using pole placement and lqr
Controller design of inverted pendulum using pole placement and lqrController design of inverted pendulum using pole placement and lqr
Controller design of inverted pendulum using pole placement and lqr
eSAT Journals
 
Lec1 01
Lec1 01Lec1 01
Sc ppt
Sc pptSc ppt
Small amplitude oscillations
Small amplitude oscillationsSmall amplitude oscillations
Small amplitude oscillations
harshsharma5537
 
Active Controller Design for Regulating the Output of the Sprott-P System
Active Controller Design for Regulating the Output of the Sprott-P SystemActive Controller Design for Regulating the Output of the Sprott-P System
Active Controller Design for Regulating the Output of the Sprott-P System
ijccmsjournal
 

Similar to Projekt (20)

Z Transform And Inverse Z Transform - Signal And Systems
Z Transform And Inverse Z Transform - Signal And SystemsZ Transform And Inverse Z Transform - Signal And Systems
Z Transform And Inverse Z Transform - Signal And Systems
 
The feedback-control-for-distributed-systems
The feedback-control-for-distributed-systemsThe feedback-control-for-distributed-systems
The feedback-control-for-distributed-systems
 
Reachability Analysis Control of Non-Linear Dynamical Systems
Reachability Analysis Control of Non-Linear Dynamical SystemsReachability Analysis Control of Non-Linear Dynamical Systems
Reachability Analysis Control of Non-Linear Dynamical Systems
 
Reachability Analysis "Control Of Dynamical Non-Linear Systems"
Reachability Analysis "Control Of Dynamical Non-Linear Systems" Reachability Analysis "Control Of Dynamical Non-Linear Systems"
Reachability Analysis "Control Of Dynamical Non-Linear Systems"
 
APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...
APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...
APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...
 
APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...
APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...
APPROXIMATE CONTROLLABILITY RESULTS FOR IMPULSIVE LINEAR FUZZY STOCHASTIC DIF...
 
Ph 101-9 QUANTUM MACHANICS
Ph 101-9 QUANTUM MACHANICSPh 101-9 QUANTUM MACHANICS
Ph 101-9 QUANTUM MACHANICS
 
Fault tolerant process control
Fault tolerant process controlFault tolerant process control
Fault tolerant process control
 
Notes nyquist plot and stability criteria
Notes nyquist plot and stability criteriaNotes nyquist plot and stability criteria
Notes nyquist plot and stability criteria
 
DSP_FOEHU - MATLAB 02 - The Discrete-time Fourier Analysis
DSP_FOEHU - MATLAB 02 - The Discrete-time Fourier AnalysisDSP_FOEHU - MATLAB 02 - The Discrete-time Fourier Analysis
DSP_FOEHU - MATLAB 02 - The Discrete-time Fourier Analysis
 
Adaptive dynamic programming algorithm for uncertain nonlinear switched systems
Adaptive dynamic programming algorithm for uncertain nonlinear switched systemsAdaptive dynamic programming algorithm for uncertain nonlinear switched systems
Adaptive dynamic programming algorithm for uncertain nonlinear switched systems
 
Controller design of inverted pendulum using pole placement and lqr
Controller design of inverted pendulum using pole placement and lqrController design of inverted pendulum using pole placement and lqr
Controller design of inverted pendulum using pole placement and lqr
 
Controller design of inverted pendulum using pole placement and lqr
Controller design of inverted pendulum using pole placement and lqrController design of inverted pendulum using pole placement and lqr
Controller design of inverted pendulum using pole placement and lqr
 
Lec1 01
Lec1 01Lec1 01
Lec1 01
 
Sc ppt
Sc pptSc ppt
Sc ppt
 
Small amplitude oscillations
Small amplitude oscillationsSmall amplitude oscillations
Small amplitude oscillations
 
Active Controller Design for Regulating the Output of the Sprott-P System
Active Controller Design for Regulating the Output of the Sprott-P SystemActive Controller Design for Regulating the Output of the Sprott-P System
Active Controller Design for Regulating the Output of the Sprott-P System
 
Rdnd2008
Rdnd2008Rdnd2008
Rdnd2008
 
Adc
AdcAdc
Adc
 
Instantons in 1D QM
Instantons in 1D QMInstantons in 1D QM
Instantons in 1D QM
 

Recently uploaded

By Design, not by Accident - Agile Venture Bolzano 2024
By Design, not by Accident - Agile Venture Bolzano 2024By Design, not by Accident - Agile Venture Bolzano 2024
By Design, not by Accident - Agile Venture Bolzano 2024
Pierluigi Pugliese
 
Enchancing adoption of Open Source Libraries. A case study on Albumentations.AI
Enchancing adoption of Open Source Libraries. A case study on Albumentations.AIEnchancing adoption of Open Source Libraries. A case study on Albumentations.AI
Enchancing adoption of Open Source Libraries. A case study on Albumentations.AI
Vladimir Iglovikov, Ph.D.
 
Uni Systems Copilot event_05062024_C.Vlachos.pdf
Uni Systems Copilot event_05062024_C.Vlachos.pdfUni Systems Copilot event_05062024_C.Vlachos.pdf
Uni Systems Copilot event_05062024_C.Vlachos.pdf
Uni Systems S.M.S.A.
 
Securing your Kubernetes cluster_ a step-by-step guide to success !
Securing your Kubernetes cluster_ a step-by-step guide to success !Securing your Kubernetes cluster_ a step-by-step guide to success !
Securing your Kubernetes cluster_ a step-by-step guide to success !
KatiaHIMEUR1
 
Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...
Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...
Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...
James Anderson
 
Monitoring Java Application Security with JDK Tools and JFR Events
Monitoring Java Application Security with JDK Tools and JFR EventsMonitoring Java Application Security with JDK Tools and JFR Events
Monitoring Java Application Security with JDK Tools and JFR Events
Ana-Maria Mihalceanu
 
Large Language Model (LLM) and it’s Geospatial Applications
Large Language Model (LLM) and it’s Geospatial ApplicationsLarge Language Model (LLM) and it’s Geospatial Applications
Large Language Model (LLM) and it’s Geospatial Applications
Rohit Gautam
 
Pushing the limits of ePRTC: 100ns holdover for 100 days
Pushing the limits of ePRTC: 100ns holdover for 100 daysPushing the limits of ePRTC: 100ns holdover for 100 days
Pushing the limits of ePRTC: 100ns holdover for 100 days
Adtran
 
DevOps and Testing slides at DASA Connect
DevOps and Testing slides at DASA ConnectDevOps and Testing slides at DASA Connect
DevOps and Testing slides at DASA Connect
Kari Kakkonen
 
みなさんこんにちはこれ何文字まで入るの?40文字以下不可とか本当に意味わからないけどこれ限界文字数書いてないからマジでやばい文字数いけるんじゃないの?えこ...
みなさんこんにちはこれ何文字まで入るの?40文字以下不可とか本当に意味わからないけどこれ限界文字数書いてないからマジでやばい文字数いけるんじゃないの?えこ...みなさんこんにちはこれ何文字まで入るの?40文字以下不可とか本当に意味わからないけどこれ限界文字数書いてないからマジでやばい文字数いけるんじゃないの?えこ...
みなさんこんにちはこれ何文字まで入るの?40文字以下不可とか本当に意味わからないけどこれ限界文字数書いてないからマジでやばい文字数いけるんじゃないの?えこ...
名前 です男
 
Essentials of Automations: The Art of Triggers and Actions in FME
Essentials of Automations: The Art of Triggers and Actions in FMEEssentials of Automations: The Art of Triggers and Actions in FME
Essentials of Automations: The Art of Triggers and Actions in FME
Safe Software
 
Introduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - CybersecurityIntroduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - Cybersecurity
mikeeftimakis1
 
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Albert Hoitingh
 
Microsoft - Power Platform_G.Aspiotis.pdf
Microsoft - Power Platform_G.Aspiotis.pdfMicrosoft - Power Platform_G.Aspiotis.pdf
Microsoft - Power Platform_G.Aspiotis.pdf
Uni Systems S.M.S.A.
 
Climate Impact of Software Testing at Nordic Testing Days
Climate Impact of Software Testing at Nordic Testing DaysClimate Impact of Software Testing at Nordic Testing Days
Climate Impact of Software Testing at Nordic Testing Days
Kari Kakkonen
 
Mind map of terminologies used in context of Generative AI
Mind map of terminologies used in context of Generative AIMind map of terminologies used in context of Generative AI
Mind map of terminologies used in context of Generative AI
Kumud Singh
 
PCI PIN Basics Webinar from the Controlcase Team
PCI PIN Basics Webinar from the Controlcase TeamPCI PIN Basics Webinar from the Controlcase Team
PCI PIN Basics Webinar from the Controlcase Team
ControlCase
 
20240607 QFM018 Elixir Reading List May 2024
20240607 QFM018 Elixir Reading List May 202420240607 QFM018 Elixir Reading List May 2024
20240607 QFM018 Elixir Reading List May 2024
Matthew Sinclair
 
A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...
sonjaschweigert1
 
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...
James Anderson
 

Recently uploaded (20)

By Design, not by Accident - Agile Venture Bolzano 2024
By Design, not by Accident - Agile Venture Bolzano 2024By Design, not by Accident - Agile Venture Bolzano 2024
By Design, not by Accident - Agile Venture Bolzano 2024
 
Enchancing adoption of Open Source Libraries. A case study on Albumentations.AI
Enchancing adoption of Open Source Libraries. A case study on Albumentations.AIEnchancing adoption of Open Source Libraries. A case study on Albumentations.AI
Enchancing adoption of Open Source Libraries. A case study on Albumentations.AI
 
Uni Systems Copilot event_05062024_C.Vlachos.pdf
Uni Systems Copilot event_05062024_C.Vlachos.pdfUni Systems Copilot event_05062024_C.Vlachos.pdf
Uni Systems Copilot event_05062024_C.Vlachos.pdf
 
Securing your Kubernetes cluster_ a step-by-step guide to success !
Securing your Kubernetes cluster_ a step-by-step guide to success !Securing your Kubernetes cluster_ a step-by-step guide to success !
Securing your Kubernetes cluster_ a step-by-step guide to success !
 
Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...
Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...
Alt. GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using ...
 
Monitoring Java Application Security with JDK Tools and JFR Events
Monitoring Java Application Security with JDK Tools and JFR EventsMonitoring Java Application Security with JDK Tools and JFR Events
Monitoring Java Application Security with JDK Tools and JFR Events
 
Large Language Model (LLM) and it’s Geospatial Applications
Large Language Model (LLM) and it’s Geospatial ApplicationsLarge Language Model (LLM) and it’s Geospatial Applications
Large Language Model (LLM) and it’s Geospatial Applications
 
Pushing the limits of ePRTC: 100ns holdover for 100 days
Pushing the limits of ePRTC: 100ns holdover for 100 daysPushing the limits of ePRTC: 100ns holdover for 100 days
Pushing the limits of ePRTC: 100ns holdover for 100 days
 
DevOps and Testing slides at DASA Connect
DevOps and Testing slides at DASA ConnectDevOps and Testing slides at DASA Connect
DevOps and Testing slides at DASA Connect
 
みなさんこんにちはこれ何文字まで入るの?40文字以下不可とか本当に意味わからないけどこれ限界文字数書いてないからマジでやばい文字数いけるんじゃないの?えこ...
みなさんこんにちはこれ何文字まで入るの?40文字以下不可とか本当に意味わからないけどこれ限界文字数書いてないからマジでやばい文字数いけるんじゃないの?えこ...みなさんこんにちはこれ何文字まで入るの?40文字以下不可とか本当に意味わからないけどこれ限界文字数書いてないからマジでやばい文字数いけるんじゃないの?えこ...
みなさんこんにちはこれ何文字まで入るの?40文字以下不可とか本当に意味わからないけどこれ限界文字数書いてないからマジでやばい文字数いけるんじゃないの?えこ...
 
Essentials of Automations: The Art of Triggers and Actions in FME
Essentials of Automations: The Art of Triggers and Actions in FMEEssentials of Automations: The Art of Triggers and Actions in FME
Essentials of Automations: The Art of Triggers and Actions in FME
 
Introduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - CybersecurityIntroduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - Cybersecurity
 
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
 
Microsoft - Power Platform_G.Aspiotis.pdf
Microsoft - Power Platform_G.Aspiotis.pdfMicrosoft - Power Platform_G.Aspiotis.pdf
Microsoft - Power Platform_G.Aspiotis.pdf
 
Climate Impact of Software Testing at Nordic Testing Days
Climate Impact of Software Testing at Nordic Testing DaysClimate Impact of Software Testing at Nordic Testing Days
Climate Impact of Software Testing at Nordic Testing Days
 
Mind map of terminologies used in context of Generative AI
Mind map of terminologies used in context of Generative AIMind map of terminologies used in context of Generative AI
Mind map of terminologies used in context of Generative AI
 
PCI PIN Basics Webinar from the Controlcase Team
PCI PIN Basics Webinar from the Controlcase TeamPCI PIN Basics Webinar from the Controlcase Team
PCI PIN Basics Webinar from the Controlcase Team
 
20240607 QFM018 Elixir Reading List May 2024
20240607 QFM018 Elixir Reading List May 202420240607 QFM018 Elixir Reading List May 2024
20240607 QFM018 Elixir Reading List May 2024
 
A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...A tale of scale & speed: How the US Navy is enabling software delivery from l...
A tale of scale & speed: How the US Navy is enabling software delivery from l...
 
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...
GDG Cloud Southlake #33: Boule & Rebala: Effective AppSec in SDLC using Deplo...
 

Projekt

  • 1. Continuous Finite-Time Stabilization of the Translational and Rotational Double Integrator Olalekan P. Ogunmolu April 30, 2015
  • 2. 1 Introduction Feedback linearization often generates closed-loop Lipschitzian dynamics Convergence in such systems is often exponential, carrying the burden of infinite settling time Undesirable in time-critical applications such as minimum-energy control, conservation of momentum et cetera It is therefore imperative to design continuous finite-time stabilizing controllers Classical optimal control provides examples of systems that converge to the equilibrium in finite time such as the double integrator. 2 The case for continuous finite-time stabilizing (FTS) con- trollers? Optimal synthesis methods such as the bang-bang time-optimal feedback control of the double integrator has discontinuous dynamics which leads to chattering or excitation of HF dynamics in flexible structures applications [Fuller et. al., ’66] The double integrator is open-loop controllable such that its states can be driven to the origin in finite time The minimum-energy control is one such method which drives the state of the system ¨x = u from an initial condition x(0) = x0, ˙x(0) = y0 to the origin in a given time tf [Athans, ’66] But such open-loop methods invite insensitivity to uncertainties and have poor disturbance rejection
  • 3. Non-Lipschitzian Dynamics of a Continuous FTS Feedback Controller 1 Overview of Problem The design of FTS continuous time-invariant feedback controllers involve non-Lipschitzian closed-loop dynamics Such controllers will exhibit non-unique solutions in backward time, i.e., better robustness and good disturbance rejection Such non-unique (revert time) solutions would violate uniqueness conditions for Lipschitz differential equations 2 Statement of Problem Consider a rigid body rotating under the action of a mechanical torque about a fixed axis Its equations of motion resemble those of a double integrator. States differ by 2nπ (where n = 0, 1, 2, . . .) in angular modes which correspond to the same physical configuration of the body. State space for this system is S1 × R rather than R2 [Andronov et. al] Developing stabilizing controls for the double integrator on R2 (translational double integrator) will lead to unwinding since the configuration space is actually R This makes an interesting problem when designing feedback controllers for the rotational double integrator with anti-wind-up compensation Discontinuous feedback controllers are practically infeasible due to the chattering they introduce because of plant uncertainties They could also excite high-frequency dynamics when used in controlling lightly damped structures [Baruh et. al.]
  • 4. 1 Finite-Time Stabilization: A Definition For the System of differential equations, ˙y(t) = f(y(t) (1) where f : D → Rn is continuous on an open neighborhood D ⊆ Rn of the origin and f(0) = 0. A continuously differentiable function y : I → D is said to be a solution of 1 on the interval I ⊂ R if y satisfies (1) for all t ∈ I We assume (1) possesses unique solutions in forward time except possibly at the origin for all initial conditions Uniqueness in forward time and the continuity of f ensure that solutions are continuous functions of initial conditions even when f is no longer Lipschitz continuous [Hartman et. al., ’82, Th. 2.1, p. 94] 2 Definition The origin is finite-time stable if there exists an open neighborhood N ⊆ D of the origin and a settling time function T : N 0 → (0, ∞), such that we have the following: 1. Finite-time convergence: For every x ∈ N {0}, ρt (x) is defined for t ∈ 0, T(x) , ρt (x) ∈ N {0}, for t ∈ 0, T(x) , and limt→T(x)ρt (x) = 0 2. Lyapunov stability: For every open set U such that 0 ∈ U ⊆ N , there exists an open set Uδ such that 0 ∈ Uδ ⊆ N and for every x ∈ Uδ {0}, ρt (x) ∈ U• for t ∈ 0, T(x) . When D = N = Rn , we have global finite-time convergence. Theorem: For a continuously differentiable function V : D → R, such that k > 0, α ∈ (0, 1) , where α and k ∈ R if there exists a neighborhood of the origin U ⊂ D such that V is positive definite, ˙V is negative definite and ˙V + kVα is negative semi-definite on U, where ˙V(x) = ∂V ∂x (x)f(x). Then, the origin of (1) is finite-time stable. Also, the settling time, T(x) is defined as T(x) = 1 k(1 − α) V(x)1−α
  • 5. 1 Continuous Finite-Time Stabilizing Controllers We want to find a continuous feedback law, u = ψ(x, y) such that the double integrator defined as, ˙x = y, ˙y = u is finite-time stabilized. 2 Proposition 1 The origin of the double integrator is globally finite-time stable [Bhat et. al., ’98, §III] under the feedback control law u where ψ(x, y) = −sign (y)|y| α − sign φα (x, y) φα (x, y) α 2 − α (2) where φα(x, y) x + 1 2 − α sign(y)|y| α 2 − α See Appendix A. for proof. 3 Remarks The vector field obtained by using the feedback control law u is locally Lipschitz everywhere except the x-axis (denoted Γ ), and the zero-level set S = {(x, y) : φα(x, y) = 0} of the function φα. The closed-loop vector field fα is transversal to Γ at every point in Γ {0, 0} Every initial condition in Γ {0, 0} has a unique solution in forward time The set S is positively invariant for the closed-loop system On the set S the closed-loop system is ˙x = −sign (x) (2 − α)|x| 1 2−α (3) ˙y = −sign (y)|y| α (4) The resulting closed loop system (4) is locally Lipschitz everywhere except the origin and therefore possesses unique solutions in forward time for initial conditions in S {0, 0}.
  • 6. 1 Example 1: Implementation of the proposed controller By choosing α = 2 3 in (2), we have the phase portrait shown in Figure 6 for the resulting feedback law ψ(x, y) = −y 2 3 − x + 3 4 y 4 3 1 2 (5) We see that all trajectories converge to the set S = {(x, y) : x + 3 4 y 4 3 = 0} in finite-time. The term −y 2 3 in (5) makes the set S positively invariant while the other term − x + 3 4 y 4 3 1 2 drives the states to S in finite-time. Therefore, (2) represents an example of a terminal sliding mode control without using discontinuous or high gain feedback.
  • 7. 1 Bounded Continuous Finite-Time Controllers In the previous section, the designed feedback controller is unbounded, meaning the controller will lead to the so-called ”unwinding” phenomenon. In a spacecraft application, for example, such unwinding can lead to the mismanagement of fuel and and momentum-consuming devices In order to finite-time stabilize the controller, we saturate the elements of the controller given in (2) It follows that for a positive number ε, satε (y) = y, |y| < ε (6) = ε sign (y) , |y| ≥ ε such that satε (y) ≤ ε for all y ∈ R (7) The relation in (7) ensures that the controller does not operate in the nonlinear region by bounding its operating range within the defined perimeter of operation (6) 1 Proposition 2 The origin of the double integrator system is a globally stable equilibrium under the bounded feedback control law u = ψ (x, y) with ψsat (x, y) = −satε {sign (y)|y| α } − {satε sign φα (x, y) φα (x, y) α 2 − α } (8) for every α ∈ (0, 1) and ε > 0, where φα(x, y) x + 1 2 − α sign(y)|y| α 2 − α See Proof in [Bhat et. al., ’98, §IV, p. 680].
  • 8. 1 Example 2: Bounded continuous finite time controller For the double integrator under the feedback control law ψsat (x, ) = −sat1(y 1 3 ) − sat1{(x + 3 5 y 5 3 )} 1 5 (9) obtained from (2) with α = 1 3 and ε = 1. Again, all trajectories converge to the set S = {(x, y) : x + 3 5 y 5 3 = 0}. But in some phase plane regions, ψsat (x, ) = 0.
  • 9. 1 The Rotational Double Integrator Let us denote the motion of a rigid body rotating about a fixed axis with unit moment of inertia as ¨θ(t) = u(t) (10) where θ is the angular displacement from some reference and u is the applied control. We can rewrite the equation as a first-order equation with ˙x = θ and ˙y = u. Suppose we require the angular position to be finite-time stable, then the feedback law given in (2) can only finite-time stabilize the origin such that if applied to the rotational double integrator, it leads to the unwinding phenomenon. Feedback controllers designed for the translational double integrator does not suffice for the rotational double integrator How to avoid the unwinding phenomenon? Modify (2) such that it is periodic in x with period 2π, i.e., ψrot (x, y) = −sign (y)|y| α − sign (sin(φα(x, y))) sin(φα(x, y)) α 2−α (11) φα is same as defined in Proposition 1. For α = 1 3 with u = ψrot the resulting phase portrait is shown below
  • 10. 1 Remarks from Phase Portrait of Rotational Double Integrator The closed loop system has equilibrium points at sn = (2nπ, 0), un = ((2n + 1)π, 0), n = · · · , −1, 0, 1, · · · The equilibrium points sn are locallu finite-time stable in forward time, while the points un are finite-time saddles The domain of attraction of the equilibrium point s is Dn = {(x, y) : (2n − 1)π < φα(x, y) < (2n + 1)π}. The shaded region of the plot shows a portion of D0. The sets Un−1 and Un represent the stable manifolds of the equilibrium points un−1 and un respectively where Un = {(x, y) : φα(x, y) = (2n + 1)π} and n = · · · , −1, 0, 1, · · · , All trajectories starting in the set Dn converge to the set Sn = {(x, y) : φα(x, y) = 2nπ} in finite, forward time and to the set Un−1 ∪ Un in finite, reverse time The sets Sn are positively invariant while the sets Un are negatively invariant 2 The solutions in the figure have no uniqueness to initial conditions lying in any of the sets Un, n = · · · , −1, 0, 1, · · ·. All solutions initialized in un are equivalent to the rigid body resting in an unstable configuration and then starting to move spontaneously clockwise or counterclockwise Departure from the unstable equilibrium is a unique feature to non-Lipschitzian systems as Lipschitzian systems do not possess solutions that depart from equilibrium The desired final configuration is however not globally finite-time stable due to the presence of the unstable equilibrium configuration at θ = π. These are saddle points un, n = · · · , −1, 0, 1, · · · This is a basic drawback to every continuous feedback controller that stabilizes the rotational double integrator without generating the unwinding effect. The desired final configuration in the phase plane corresponds to multiple equilibria in the phase plane meaning every controller that stabilizes the desired configuration stabilizes each equilibria But stability, continuous dependence on initial conditions and solutions’ uniqueness imply that the domain of attraction of any two equilibrium points in the plane are non-empty, open and disjoint.
  • 11. 1 We cannot write R2 as the union of a collection of disjoint sets. Thus, there are initial conditions in the plane that do not converge to the equilibria of the desired final configuration. With respect to (11), these initial conditions are the stable manifolds of the unstable configuration. The designed controller is practically globally stable as its non-Lipschitzian property increases the sensitivity of the unstable configuration to perturbations 1 Appendix A: Proof of Proposition 1 If we denote φα(x, y) by φα and fix α ∈ (0, 1), we could choose the C2 Lyapunov function candidate, V (x, y) = 2 − α 3 − α |φα| 3 − α 2 − α + syφα + r 3 − α |y| 3−α (12) where r and s are positive numbers. Along the closed loop trajectories, ˙V (x, y) = sφα ˙y + r|y| 2−α ˙y + sy ˙φα +|φα| 1 2 − α ˙φα = sφα    −sign (y)|y| α − sign (φα)|φα| α 2 − α     + r|y| 2−α     (13) +|φα| 1 2 − α ˙φα But, ˙φα = ˙x + ˙y sign (y) |y| 1−α (14) from (2), it therefore follows that,
  • 12. 1 Continuation of Proof of Proposition 1 ˙φα = −sign (y) sign (φα)|y|1−α |φα| α 2−α (15) Putting (14) into (13), and noting that sy ˙φα = −s sign (yφα)|y| 1−α |φα| 1 + α 2 − α (16) and ˙φα|φα| 1 2−α = −sign (y) sign (φα) |y| 1−α |φα| 1 + α 2 − α (17) we find that, (18)˙V (x, y) = −ry 2 − s|φα| 2 2−α −|y| 1−α |φα| 1+α 2−α − sφαsign (y)|y| α − (r + s)sign(yφα)|y| 2−α |φα| α 2−α 1 Remarks The obtained Lyapunov derivative in the foregoing is continuous everywhere since α ∈ (0, 1). For k > 0 and (x, y) ∈ R2 the following holds, Introduce x = k2−α , y = ky such that
  • 13. 1 Remarks (19) φα(k 2−α x, ky) = k 2−α x − 1 2 − α sign(ky)|ky| 2−α = k 2−α φα(x, y)) and (20)V(k 2−α x, ky) = 2 − α 3 − α φα(k 2−α x, ky) 3 − α 2 − α + skyφα(k 2−α x, ky) + r 3 − α |ky| 3−α such that (21)V(k 2−α x, ky) = k 3−α V(x, y) Following a similar logic as in (20), we find that (22)˙V(k 2−α x, ky) = k 2 ˙V(x, y) The results of the previous section imply that for r > 1 and s < 1, both V and ˙V are positive on the set O = {(x, y) : max(x,y)=(0,0)|φα| 1 2−α ,|y| = 1} which is a closed curve encircling the origin. For every (x, y) ∈ R2 {0, 0} there exists k > 0 such that k2−α x, ky ∈ O, the homogeneity properties of (20) and (22) imply a positive definite V and negative semi-definite ˙V. From (20), V is radially unbounded so that the set V = {(x, y) : V(x, y) = 1} is compact. Therefore ˙V achieves its maximum on the compact set V. If we define c = −max{(x,y)∈V} ˙V(x, y), then ˙V(x, y) ≤ −c{V(x, y)} 2 3−α for all (x, y) ∈ R2 [Bhat et. al. ’96] The homogeneity of (20) and (22) ensures ˙V(x, y) ≤ −cV(x, y) 2 3−α for all (x, y) ∈ R2 [Bhat et. al. ’96]. Since α ∈ (0, 1) is ≡ 2 3−α ∈ (0, 1), we can conclude finite time stability
  • 14. 1 References A. A. Andronov, A. A. Vitt, and S. E. Khaikin. Theory of Oscillators. Oxford, UK: Pergamon, 1966. M. Athans and P.L. Falb. Optimal Control: An Introduction to the Theory and Its Applications. McGraw-Hill, New York, 1966. H. Baruh and S.S.K. Tadikonda. Gibbs phenomenon in structural control. J. Guidance, Contr., and Dynamics. vol., no. 1, pp. 51-58, 1991. S.P. Bhat and D.S. Bernstein. Continuous Finite-Time Stabilization of the Translational and Rotational Double Integra- tors. IEEE Transactions on Automatic Control. vol. 43., no. 5, 1998 S.P. Bhat and D.S. Bernstein. Continuous, Bounded, Finite-Time Stabilization of the Translational and Rotational Double Integrators. in Proc. Amer, Contr. Conf., Seattle, WA, June 1996, pp. 1831 - 1832 A. Fuller. Optimization of some nonlinear control systems by means of Bellman’s equation and dimensional analysis. Int. J. Contr., vol. 3, no. 4, pp. 359–394, 1966. P. Hartman. Ordinary Differential Equations. 2nd edition, Boston, MA: Birkhauser, 1982.