Scribd
Upload
9 views2 pages

Luenberger Observers

Luenberger observers are used in control systems to estimate the state when direct measurement is not possible, utilizing a separation theorem that incorporates Kalman-Bucy filters. The document discusses how to construct a controller by estimating the state through a cascade of linear time-invariant systems and highlights the relationship between the actual state and the surrogate state. It also details the mathematical framework for recovering the state vector using known inputs and outputs, ensuring stability in the estimation process.

Uploaded by

mnaghibian80
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
9 views2 pages

Luenberger Observers

Luenberger observers are used in control systems to estimate the state when direct measurement is not possible, utilizing a separation theorem that incorporates Kalman-Bucy filters. The document discusses how to construct a controller by estimating the state through a cascade of linear time-invariant systems and highlights the relationship between the actual state and the surrogate state. It also details the mathematical framework for recovering the state vector using known inputs and outputs, ensuring stability in the estimation process.

Uploaded by

mnaghibian80
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

8

Luenberger Observers

Linear regulator controller designs result in a state variable feedback law, so that
implementing an optimal control requires measurement of all components of the
state. In many situations such measurements are not possible and alternative ap-
proaches are needed. One such approach was originated by D. G. Luenberger, and
is known as Luenberger observers.
The separation theorem controllers (see Chapter 7) use a Kalman-Bucy filter to
estimate the state, and construct the controller by feeding back the optimal gains
applied to the estimate rather than the (unavailable) actual state.
Construction of a separation theorem controller requires specifying (estimating)
models for observation and disturbance noise processes in order to compute the
estimation filter gains. Some systems are relatively "noiseless", so it might be
thought that the benefits of the separation theorem controller structure could be
attained without investing in full stochastic models.!
If the problem is approached strictly as one of discovering a "surrogate state"
rather than an optimal state estimate, then some insight can be gained just from
the structure of the cascade of linear time-invariant systems.
Suppose that the state is governed by
dx
- =Ax+ Bu, y=Cx,
dt
and that the system is assumed observable. Then knowing uO and y(.) on an
interval suffices to determine x(O) and hence the state vector (at least with some
small delay).

1Specifying models with small observation noise lead~ to "aggressive" filters with fast dynamics.

J. H. Davis, Foundations of Deterministic and Stochastic Control


© Birkhäuser Boston 2002
246 8. Luenberger Observers

If the output of this system is used as the input to a second system in the form
dz
- =Fz+ Gy,
dt
then the response of this "filter equation" is

Z(s) = (Is - F)-lzO + (Is - F)-lGC(Is - A)-lxO


+ (Is - F)-lGC(Is - A)-lU(s).

Since the filter initial condition zo and the input uO are assumed known, this
suggests that x can be recovered as a linear transformation of z within transient
and measurable terms.
To verify this, consider first the case of a zero input u(·). Then
dx
-=Ax,
dt
and a linear transformation on the state satisfies
dx
T- =TAx.
dt
The difference between Tx and the surrogate state z satisfies
d
- (z - Tx) = -TAx +Fz + GCx
dt .
= - T A x + F (z - Tx) + GCx + FTx
= F(z - Tx) + (FT - TA + GC) x.
Provided that the matrices involved satisfy

(FT - TA + GC) = 0,
then
d
- (z - Tx) = F (z - Tx)
dt
and z approaches Tx as long as F is chosen stable.
These manipulations carry over to the situation where an input function uO is
present. Then
dx
T- = TAx + TBu(t) ,
dt
and the "estimator" equation simply has to be modified to

-dz
dt
= F z + Gy + TBu
in order to obtain the same error equation
d
- (z - Tx) = F (z - Tx) .
dt

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy