• Tidak ada hasil yang ditemukan

Separable Subsystem Control

Chapter IV: Separable Subsystems

4.2 Separable Subsystem Control

Constructing the feedback linearizing controller in (3.6) requires the dynamics of the full-order system. However, in the case of large dimensional systems, the full dynamics may be unknown or may become computationally expensive, inhibiting feedback linearization.

Figure 4.1: Amputee-Prosthesis Separable System and Equivalent Subsystem.

(Left) Amputee-prosthesis separable system (blue), with separable prosthesis sub- system AMPRO3 (red). (Right) Equivalent prosthesis subsystem. (Middle) Control input from inverse dynamics of human-prosthesis motion capture walking data, determined with full-order system dynamics (blue) and with equivalent subsystem dynamics (red).

Control Law for Separable Subsystem

This section eliminates the need to know the full-order system dynamics for feedback linearization by constructing aseparable subsystem control law that only depends on subsystem dynamics. We begin by defining aseparable control system.

Definition 1: The affine control system(3.1)is aseparable control systemif it can be structured as

"

ยค ๐‘ฅ๐‘Ÿ

ยค ๐‘ฅ๐‘ 

#

=

"

๐‘“๐‘Ÿ(๐‘ฅ) ๐‘“๐‘ (๐‘ฅ)

# +

"

๐‘”๐‘Ÿ

1(๐‘ฅ) ๐‘”๐‘Ÿ

2(๐‘ฅ) 0 ๐‘”๐‘ (๐‘ฅ)

# "

๐‘ข๐‘Ÿ ๐‘ข๐‘ 

# , ๐‘ฅ๐‘Ÿ โˆˆR๐‘›๐‘Ÿ, ๐‘ฅ๐‘  โˆˆR๐‘›๐‘ , ๐‘ข๐‘Ÿ โˆˆR๐‘š๐‘Ÿ, ๐‘ข๐‘  โˆˆR๐‘š๐‘ ,

(4.1)

where๐‘›๐‘Ÿ +๐‘›๐‘  =๐‘›and๐‘š๐‘Ÿ +๐‘š๐‘  =๐‘š.

Because of the structure of๐‘”(๐‘ฅ) in (4.1), ๐‘ข๐‘Ÿ only acts on part of the system. This motivates defining aseparable subsystemindependent of๐‘ข๐‘Ÿ.

Definition 2: For a separable control system (4.1), its separable subsystem is defined as

ยค

๐‘ฅ๐‘  = ๐‘“๐‘ (๐‘ฅ) +๐‘”๐‘ (๐‘ฅ)๐‘ข๐‘ , (4.2) which depends on the full-order system states๐‘ฅ โˆˆR๐‘›.

Now, to construct a feedback linearizing control law for this separable subsystem, we construct output functions that solely depend on the subsystem states๐‘ฅ๐‘  โˆˆ R๐‘›๐‘  and whose Lie derivatives solely depend on the subsystem (4.2).

Definition 3: For a separable subsystem (4.2) of the separable control system (4.1), a set of linearly independent output functions with vector relative degree

ยฎ

๐›พ๐‘  =(๐›พ๐‘ 

1, ๐›พ๐‘ 

2, . . . , ๐›พ๐‘ 

๐‘š๐‘ )with respect to(4.1)areseparable subsystem outputsif they only depend on๐‘ฅ๐‘  โˆˆR๐‘›๐‘ ,

๐‘ฆ๐‘ (๐‘ฅ๐‘ ) โˆˆR๐‘š๐‘ , (4.3)

and meet the following cross-term cancellation conditions for ๐‘— =1, . . . , ๐›พ๐‘ 

๐‘– โˆ’1and ๐‘–=1, . . . , ๐‘š๐‘ :

๐œ• ๐ฟ

๐‘— ๐‘“๐‘ ๐‘ฆ๐‘ (๐‘ฅ)

๐œ• ๐‘ฅ๐‘Ÿ

๐‘“๐‘Ÿ(๐‘ฅ) =0, (D3.1)

๐œ• ๐ฟ

๐›พ๐‘ 

๐‘–โˆ’1 ๐‘“๐‘ 

๐‘ฆ๐‘ (๐‘ฅ)

๐œ• ๐‘ฅ๐‘Ÿ h

๐‘”๐‘Ÿ

1(๐‘ฅ) ๐‘”๐‘Ÿ

2(๐‘ฅ)i

= h 0 0

i

. (D3.2)

We use these outputs to introduce aseparable subsystem control lawin terms of the subsystem (4.2) alone.

Definition 4: For a separable subsystem (4.2) with separable subsystem outputs (4.3), we define a separable subsystem control law as the feedback linearizing control law

๐‘ขssc(๐‘ฅ) โ‰œ โˆ’(๐ฟ๐‘”๐‘ ๐ฟยฎ

๐›พ๐‘ โˆ’1 ๐‘“๐‘ 

๐‘ฆ๐‘ (๐‘ฅ)

| {z }

๐ด๐‘ (๐‘ฅ)

)โˆ’1(๐ฟยฎ

๐›พ๐‘  ๐‘“๐‘ 

๐‘ฆ๐‘ (๐‘ฅ)

| {z }

๐ฟโˆ—

๐‘“๐‘ ๐‘ฆ๐‘ (๐‘ฅ)

โˆ’๐œ‡๐‘ )

=โˆ’๐ดโˆ’1

๐‘  (๐‘ฅ) (๐ฟโˆ—

๐‘“๐‘ ๐‘ฆ๐‘ (๐‘ฅ) โˆ’๐œ‡๐‘ ).

(4.4)

This control law is independent of the rest of the system dynamics ๐‘“๐‘Ÿ(๐‘ฅ), ๐‘”๐‘Ÿ

1(๐‘ฅ), and๐‘”๐‘Ÿ

2(๐‘ฅ), but still depends on the full-order system states๐‘ฅ. We will address this dependence in subsequent results to develop an implementable form of this control law solely dependent on subsystem states and measurable inputs.

To compare this control law๐‘ขssc(๐‘ฅ)to๐‘ข๐‘ (๐‘ฅ), we constructseparable outputsfor the full-order system that include the separable subsystem outputs๐‘ฆ๐‘ (๐‘ฅ๐‘ )used for (4.4).

Definition 5: For a separable control system, a set of linearly independent output functions with vector relative degree๐›พยฎareseparable outputsif they are structured

as

๐‘ฆ(๐‘ฅ) =

"

๐‘ฆ๐‘Ÿ(๐‘ฅ) ๐‘ฆ๐‘ (๐‘ฅ๐‘ )

#

, ๐‘ฆ๐‘Ÿ(๐‘ฅ) โˆˆR๐‘š๐‘Ÿ, ๐‘ฆ๐‘ (๐‘ฅ๐‘ ) โˆˆR๐‘š๐‘ , (4.5) and ๐‘ฆ๐‘ (๐‘ฅ๐‘ ) are separable subsystem outputs with vector relative degree ๐›พยฎ๐‘ . The remaining outputs ๐‘ฆ๐‘Ÿ(๐‘ฅ) have vector relative degree๐›พยฎ๐‘Ÿ and can depend on any of the system states๐‘ฅ. The number of subsystem outputs๐‘š๐‘  and the number of the rest of the outputs๐‘š๐‘Ÿ sums to๐‘š, and๐›พยฎ=( ยฎ๐›พ๐‘Ÿ,๐›พยฎ๐‘ ).

For the following theorem, we define the auxilary control input ๐œ‡as divided in the following form:

๐œ‡=

"

๐œ‡๐‘Ÿ ๐œ‡๐‘ 

#

, ๐œ‡๐‘Ÿ โˆˆR๐‘š๐‘Ÿ, ๐œ‡๐‘  โˆˆR๐‘š๐‘ . (4.6) Theorem 1: For a separable control system (4.1), if the control law ๐‘ข(๐‘ฅ) = (๐‘ข๐‘Ÿ(๐‘ฅ)๐‘‡, ๐‘ข๐‘ (๐‘ฅ)๐‘‡)๐‘‡ (3.6) is constructed with separable outputs (4.5) and auxiliary control input(4.6), then๐‘ข๐‘ (๐‘ฅ)=๐‘ขssc(๐‘ฅ).

Proof: We begin by relating the 3 components (๐ดโˆ’1(๐‘ฅ), ๐ฟโˆ—

๐‘“

๐‘ฆ(๐‘ฅ), ๐œ‡)of๐‘ข(๐‘ฅ)to the components of๐‘ขssc(๐‘ฅ), (๐ดโˆ’1

๐‘  (๐‘ฅ), ๐ฟโˆ—

๐‘“๐‘ 

๐‘ฆ๐‘ (๐‘ฅ), ๐œ‡๐‘ ). We are given ๐œ‡ = โ˜… ๐œ‡๐‘ 

by (4.6).

With condition (D3.2), we show:

๐ด(๐‘ฅ) = ๏ฃฎ

๏ฃฏ

๏ฃฏ

๏ฃฏ

๏ฃฏ

๏ฃฏ

๏ฃฐ

๐œ• ๐ฟยฎ

๐›พ๐‘Ÿ

โˆ’1 ๐‘“ ๐‘ฆ๐‘Ÿ(๐‘ฅ)

๐œ• ๐‘ฅ๐‘Ÿ

๐œ• ๐ฟยฎ

๐›พ๐‘Ÿ

โˆ’1 ๐‘“ ๐‘ฆ๐‘Ÿ(๐‘ฅ)

๐œ• ๐‘ฅ๐‘ 

๐œ• ๐ฟยฎ

๐›พ๐‘ 

โˆ’1 ๐‘“๐‘  ๐‘ฆ๐‘ (๐‘ฅ)

๐œ• ๐‘ฅ๐‘Ÿ

๐œ• ๐ฟยฎ

๐›พ๐‘ 

โˆ’1 ๐‘“๐‘  ๐‘ฆ๐‘ (๐‘ฅ)

๐œ• ๐‘ฅ๐‘ 

๏ฃน

๏ฃบ

๏ฃบ

๏ฃบ

๏ฃบ

๏ฃบ

๏ฃป

"

๐‘”๐‘Ÿ

1(๐‘ฅ) ๐‘”๐‘Ÿ

2(๐‘ฅ) 0 ๐‘”๐‘ (๐‘ฅ)

#

=

๏ฃฎ

๏ฃฏ

๏ฃฏ

๏ฃฏ

๏ฃฏ

๏ฃฐ

โ˜… โ˜…

0

๐œ• ๐ฟยฎ

๐›พ๐‘ 

โˆ’1 ๐‘“๐‘  ๐‘ฆ๐‘ (๐‘ฅ)

๐œ• ๐‘ฅ๐‘  ๐‘”๐‘ (๐‘ฅ)

๏ฃน

๏ฃบ

๏ฃบ

๏ฃบ

๏ฃบ

๏ฃป

=

"

โ˜… โ˜…

0 ๐ด๐‘ (๐‘ฅ)

# ,

๐ด(๐‘ฅ)โˆ’1=

"

โ˜… โ˜…

0 ๐ด๐‘ (๐‘ฅ)โˆ’1

# ,

where the final step is true because of the 0 corner block and properties of block matrix inversion. Similarly, we show:

๐ฟโˆ—

๐‘“๐‘ฆ(๐‘ฅ)=

๏ฃฎ

๏ฃฏ

๏ฃฏ

๏ฃฏ

๏ฃฏ

๏ฃฏ

๏ฃฐ

๐œ• ๐ฟยฎ

๐›พ๐‘Ÿ

โˆ’1 ๐‘“

๐‘ฆ๐‘Ÿ(๐‘ฅ)

๐œ• ๐‘ฅ๐‘Ÿ

๐œ• ๐ฟยฎ

๐›พ๐‘Ÿ

โˆ’1 ๐‘“

๐‘ฆ๐‘Ÿ(๐‘ฅ)

๐œ• ๐‘ฅ๐‘ 

๐œ• ๐ฟยฎ

๐›พ๐‘ 

โˆ’1 ๐‘“๐‘  ๐‘ฆ๐‘ (๐‘ฅ)

๐œ• ๐‘ฅ๐‘Ÿ

๐œ• ๐ฟยฎ

๐›พ๐‘ 

โˆ’1 ๐‘“๐‘  ๐‘ฆ๐‘ (๐‘ฅ)

๐œ• ๐‘ฅ๐‘ 

๏ฃน

๏ฃบ

๏ฃบ

๏ฃบ

๏ฃบ

๏ฃบ

๏ฃป

"

๐‘“๐‘Ÿ(๐‘ฅ) ๐‘“๐‘ (๐‘ฅ)

#

=

๏ฃฎ

๏ฃฏ

๏ฃฏ

๏ฃฏ

๏ฃฏ

๏ฃฐ

โ˜…

๐œ• ๐ฟยฎ

๐›พ๐‘ 

โˆ’1 ๐‘“๐‘  ๐‘ฆ๐‘ (๐‘ฅ)

๐œ• ๐‘ฅ๐‘ 

๐‘“๐‘ (๐‘ฅ)

๏ฃน

๏ฃบ

๏ฃบ

๏ฃบ

๏ฃบ

๏ฃป

=

"

โ˜… ๐ฟโˆ—

๐‘“๐‘ 

๐‘ฆ๐‘ (๐‘ฅ)

# .

Putting these components together to construct๐‘ข(๐‘ฅ) yields ๐‘ข(๐‘ฅ)=โˆ’

"

โ˜… โ˜…

0 ๐ดโˆ’1

๐‘  (๐‘ฅ)

# "

โ˜… ๐ฟโˆ—

๐‘“๐‘ 

๐‘ฆ๐‘ (๐‘ฅ)

#

โˆ’

"

โ˜… ๐œ‡๐‘ 

#

=

"

โ˜…

โˆ’๐ดโˆ’1

๐‘  (๐‘ฅ) (๐ฟโˆ—

๐‘“๐‘ ๐‘ฆ๐‘ (๐‘ฅ) โˆ’๐œ‡๐‘ )

#

=

"

โ˜… ๐‘ขssc(๐‘ฅ)

# , showing๐‘ข๐‘ (๐‘ฅ) =๐‘ขssc(๐‘ฅ) as defined in (4.4).

By Theorem 1, we can construct a stabilizing controller (4.4) for the subsystem independent of the rest of the system dynamics, identical to the portion of the controller constructed with the full-order system dynamics acting on the subsystem.

In the case of full-state feedback linearizable systems [39], this guarantees full- order system stability when (3.6) is applied to the rest of the system. This enables construction of stablemodel dependentcontrollers for separable subsystems without knowledge of the full-order system dynamics.

Equivalency of Subsystems

Although we now have a subsystem control law independent of the rest of the system dynamics, it still depends on the full-order system states๐‘ฅ โˆˆ R๐‘›. Consider the case where the states ๐‘ฅ๐‘Ÿ โˆˆ R๐‘›๐‘Ÿ cannot be measured. If we could construct an equivalent subsystem whose dynamics are a function of the subsystem states ๐‘ฅ๐‘  โˆˆR๐‘›๐‘ , and measurable inputsF โˆˆR๐‘›๐‘“, we could calculate the subsystem control lawindependent of the full-order system states.

Consider another subsystem,

ยคยฏ

๐‘ฅ๐‘  = ๐‘“ยฏ๐‘ (X) +๐‘”ยฏ๐‘ (X)๐‘ขยฏ๐‘  X =

"

ยฏ ๐‘ฅ๐‘  F

#

=

"

๐‘ฅ๐‘  F

#

โˆˆR๐‘›ยฏ,

(4.7)

where X is the state vector ยฏ๐‘ฅ๐‘  = ๐‘ฅ๐‘  augmented with an input F. Using the same separable subsystem outputsas (4.3), we define a control law ยฏ๐‘ข๐‘ (X)for the subsystem as:

ยฏ

๐‘ข๐‘ (X) โ‰œ โˆ’(๐ฟ๐‘”ยฏ๐‘ ๐ฟยฎ

๐›พ๐‘ โˆ’1

ยฏ ๐‘“๐‘ 

๐‘ฆ๐‘ (X)

| {z }

ยฏ ๐ด๐‘ (X)

)โˆ’1(๐ฟยฎ

๐›พ๐‘ 

ยฏ ๐‘“๐‘ 

๐‘ฆ๐‘ ( X)

| {z }

๐ฟโˆ—

ยฏ ๐‘“๐‘ ๐‘ฆ๐‘ (X)

โˆ’๐œ‡๐‘ )

=โˆ’๐ดยฏโˆ’1

๐‘  (X) (๐ฟโˆ—

ยฏ ๐‘“๐‘ 

๐‘ฆ๐‘ (X) โˆ’๐œ‡๐‘ ).

(4.8)

Theorem 2: For the subsystems (4.2) and(4.7), ifโˆƒ ๐‘‡ : R๐‘› โ†’ R๐‘›ยฏ s.t. ๐‘‡(๐‘ฅ) = X and the following conditions hold,

๐‘“๐‘ (๐‘ฅ) = ๐‘“ยฏ๐‘ (X), ๐‘”๐‘ (๐‘ฅ) =๐‘”ยฏ๐‘ (X),

(T2) then ๐‘ข๐‘ (๐‘ฅ) = ๐‘ขยฏ๐‘ (X). Applying these to (4.2) and (4.7), respectively, results in dynamical systems such that given the same initial condition๐‘ฅ๐‘Ÿ0

๐‘ฅ๐‘ 0

= h๐‘ฅ๐‘Ÿ(๐‘ก0)

๐‘ฅ๐‘ (๐‘ก0)

i yields solutions๐‘ฅ๐‘ (๐‘ก) =๐‘ฅยฏ๐‘ (๐‘ก) โˆ€๐‘ก โ‰ฅ ๐‘ก0.

Proof. Since the subsystems have the same dynamics and outputs, the Lie derivatives comprising their control laws are also the same, hence

๐‘ข๐‘ (๐‘ฅ) =โˆ’๐ดโˆ’1

๐‘  (๐‘ฅ) (๐ฟโˆ—

๐‘“๐‘ ๐‘ฆ๐‘ (๐‘ฅ) โˆ’๐œ‡๐‘ )

=โˆ’๐ดยฏโˆ’1

๐‘  (X) (๐ฟโˆ—

ยฏ ๐‘“๐‘ 

๐‘ฆ๐‘ (X) โˆ’๐œ‡๐‘ ) =๐‘ขยฏ๐‘ (X).

With the same control law and dynamics, the closed-loop dynamics of the subsys- tems are the same:

ยค

๐‘ฅ๐‘  = ๐‘“๐‘ (๐‘ฅ) +๐‘”๐‘ (๐‘ฅ)๐‘ข๐‘ (๐‘ฅ)

= ๐‘“ยฏ๐‘ (X) +๐‘”ยฏ๐‘ (X)๐‘ขยฏ๐‘ (X)=๐‘ฅยคยฏ๐‘ . Hence, given the same initial condition๐‘ฅ๐‘Ÿ0

๐‘ฅ๐‘ 0

= h

๐‘ฅ๐‘Ÿ(๐‘ก0) ๐‘ฅ๐‘ (๐‘ก0)

i

, they have the same solution ๐‘ฅ๐‘ (๐‘ก)=๐‘ฅยฏ๐‘ (๐‘ก) โˆ€๐‘ก โ‰ฅ ๐‘ก0.

By Theorems 1 and 2, we can construct a stabilizingmodel dependent controller, namely (4.8), for the subsystem (4.2) independent of the rest of the system dynamics and with measurable inputs.

Zero Dynamics. For full-state feedback linearizable systems, this control method will stabilize the full-order system. For partially feedback linearizable systems, we can apply this method to the feedback linearizable portion of the system. If the zero dynamics are stable, then we can guarantee full-order system stability, which will be proved through Lyapunov methods in the next chapter, Chapter 5.