Scribd
Upload
16 views20 pages

10 - State Space Methods

Uploaded by

Salim Saleh
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
16 views20 pages

10 - State Space Methods

Uploaded by

Salim Saleh
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/ 20

10.

State space analysis

• Overview
• Definition of important terms
• State space models
• Linearization
• Solution of the state equation
• Stability
• Controllability
• Observability
• Transfer function of a state-space model
• State-space Design
Overview

• State space models present an alternative approach


to modeling dynamical systems.
• Advantages of state variable analysis are: Linear
systems; Non-linear system; Time variant systems;
Time invariant systems; Multiple input multiple
output systems; This gives idea about the internal
state of the system
Definition of important terms

• State: The state of a dynamic system is the smallest


set of variables called state variables such that
the knowledge of these variables at time (𝑡 = 𝑡 0 )
(Initial condition), together with the knowledge
of input for (𝑡 ≥ 𝑡0 ), completely determines the
behavior of the system for any time (𝑡 ≥ 𝑡 0 ).
• State Variables: A set of variables which describes
the system at any time instant are called state
variables.
• State vector: A vector whose elements are the state
variables.
Definition of important terms

• State space: The n-dimensional space whose


coordinate axes consists of the 𝑥1 axis, 𝑥2 axis,....
𝑥 𝑛 axis (where 𝑥1 , 𝑥2 , .....𝑥 𝑛 are state variables:) is
called a state space.
State space models

• A state-space model of an 𝑛th order dynamical


system is defined by 𝑛 coupled first-order
differential equations:

• where 𝑥 1 to 𝑥 𝑛 are the state variables, 𝑢 is the input


variable, 𝑥¤𝑖 = 𝑑𝑥𝑖 /𝑑𝑡, 𝑡 is the time variable, and 𝑓1
to 𝑓𝑛 are continuous functions of their arguments.
State space models
• The state-space modelas the vector equation (state
equation) is given as
𝑥¤ = 𝑓 (𝑥, 𝑢)

• Then, an output equation that defines the output


variable 𝑦 as a function of the state and input
variables:
𝑦 = ℎ(𝑥, 𝑢)
State space models

• The state equation and output equation constitute


the state-space model of the system is defined as
• For 𝑛 = 3, the state equation is expressed as
State space models

• Then, the state equation is defined as


𝑥¤ = 𝐴𝑥 + 𝐵𝑢
• Similarly, for the output equation, if
ℎ = 𝑐 1 𝑥1 + 𝑐 2 𝑥2 + 𝑐 3 𝑥3 + 𝑑𝑢

• The state-space model for linear systems takes the


form
𝑥¤ = 𝐴𝑥 + 𝐵𝑢
𝑦 = 𝐶𝑥 + 𝐷𝑢
Linearization

• Nonlinear systems can be modeled by linear


models in the neighborhood of an equilibrium
point.
• Linearization of a nonlinear state-space model
uses Jacobian matrices, which are defined as
follows.
• For 𝑓 (𝑥, 𝑢), the Jacobian matrix [𝛿 𝑓 /𝛿𝑥] is
defined by
Linearization

• The Jacobian matrix [𝛿 𝑓 /𝛿𝑢] (𝑥, 𝑢) is defined by

• For the function ℎ(𝑥, 𝑢), the Jacobian matrix


[𝛿ℎ/𝛿𝑥] is defined by
Linearization

• Suppose that for a constant input 𝑥, ¯ the system


𝑥¤ = 𝑓 (𝑥, 𝑢) has an equilibrium point at 𝑥;
¯ that is,

0 = 𝑓 ( 𝑥,
¯ 𝑢)
¯
• Then, the system can be approximated in the
neighborhood of the equilibrium point by the
linear state-space model

𝑥¤ 𝛿 = 𝐴𝑥 𝛿 + 𝐵𝑢 𝛿

𝑦 𝛿 = 𝐶𝑥 𝛿 + 𝐷𝑢 𝛿
Linearization

• where 𝑥 𝛿 = 𝑥 − 𝑥,
¯ 𝑢 𝛿 = 𝑢 − 𝑢,¯ and 𝑦 𝛿 = 𝑦 − ℎ( 𝑥,
¯ 𝑢)
¯
are perturbations from their equilibrium values,
and the matrices 𝐴, 𝐵, 𝐶, 𝐷 are given by
𝛿𝑓
𝐴= ( 𝑥,
¯ 𝑢)¯
𝛿𝑥
𝛿𝑓
𝐵= ( 𝑥,
¯ 𝑢)¯
𝛿𝑢
𝛿ℎ
𝐶= ( 𝑥,
¯ 𝑢)¯
𝛿𝑥
𝛿ℎ
𝐷= ( 𝑥,
¯ 𝑢)¯
𝛿𝑢
Solution of the state equation

• The state equation

𝑥¤ = 𝐴𝑥(𝑡) + 𝐵𝑢(𝑡), 𝑢(𝑡), 𝑥(0) = 𝑥0


• Take Laplace transform

𝑋 (𝑠) = (𝑠𝐼 − 𝐴) −1 𝑥 0 + (𝑠𝐼 − 𝐴) −1 𝐵𝑈 (𝑠)


• Take inverse Laplace transform (time-domain)

L {𝑋 (𝑠)} = L (𝑠𝐼 − 𝐴) −1 𝑥 0


+L (𝑠𝐼 − 𝐴) −1 𝐵𝑈 (𝑠)

Solution of the state equation

• Define the state transition matrix Φ by

Φ = L (𝑠𝐼 − 𝐴) −1


• The solution to state equation is given by


∫ 𝑡
𝑥(𝑡) = Φ(𝑡)𝑥 0 + Φ(𝑡 − 𝜏)𝐵𝑈 (𝜏)𝑑𝜏
0

• Substitution of 𝑥(𝑡) in the output equation

𝑦(𝑡) = 𝐶𝑥(𝑡) + 𝐷𝑢(𝑡) yields


∫ 𝑡
𝑦(𝑡) = 𝐶Φ(𝑡)𝑥0 + 𝐶Φ(𝑡 − 𝜏)𝐵𝑈 (𝜏)𝑑𝜏 + 𝐷𝑢(𝑡)
0
Stability

The system characterized by (state space model)

𝑥¤ = 𝐴𝑥 + 𝐵𝑢

𝑦 = 𝐶𝑥 + 𝐷𝑢
is stable if
1. the solution of 𝑥¤ = 𝐴𝑥, with 𝑥(0) = 𝑥0 , tends to
zero as 𝑡 tends to infinity, and
2. the solution of 𝑥¤ = 𝐴𝑥 + 𝐵𝑢 is bounded for every
bounded input.
Stability

• The system characterized by state space model is


characterized by the eigenvalues of the matrix 𝐴.
• The eigenvalues of a real matrix 𝐴 are the roots of

det (𝑠𝐼 − 𝐴) = 0
• The system of state space model is stable if and only
if all eigenvalues of 𝐴 have negative real parts.
Controllability

• Controllability and observability play an


important role in the design of feedback control.
• The system 𝑥¤ = 𝐴𝑥 + 𝐵𝑢, or the pair ( 𝐴, 𝐵), is
controllable if for any initial state 𝑥0 , there is a
continuous control 𝑢(𝑡) that steers the state of the
system to any desired final state 𝑥 𝑓 over the time
period [0, 𝑡 𝑓 ], for some finite 𝑡 𝑓 .
• The 𝑛−dimensional system 𝑥¤ = 𝐴𝑥 + 𝐵𝑢 is
controllable if and only if rank 𝐶 𝑀 = 𝑛 (full
row rank), where 𝐶 𝑀 is the controllability matrix,
defined by
𝐶 𝑀 = 𝐵 𝐴𝐵 𝐴2 𝐵 ... 𝐴𝑛−1 𝐵
 
Observability

• Observability is being able


to estimate the value of (full column rank), where
the state of the system 𝑂 𝑀 is the observability
at different points in time matrix, defined by
based on the output.  𝐶 
 
• The 𝑛−dimensional system  𝐶𝐴 
 
 𝐶 𝐴2 
𝑥¤ = 𝐴𝑥
 
𝑂𝑀 =  . 
 
 
𝑦 = 𝐶𝑥  . 
 
 . 
• or the pair ( 𝐴, 𝐶), is  𝑛−1 
𝐶 𝐴 
observable if and only if  
rank 𝑂 𝑀 = 𝑛
Transfer function of a state-space model

𝐺 (𝑠) = 𝑐(𝑠𝐼 − 𝐴) −1 𝐵 + 𝐷
State-space design (Reading assignment )

• Design an output feedback controller to stabilize


the system
𝑥¤ = 𝐴𝑥 + 𝐵𝑢, 𝑦 = 𝐶𝑥
   
0 1 0  
where 𝐴 = , 𝐵= , 𝐶= 1 0
0 0 1
• Consider the system defined by
𝑥¤ = 𝐴𝑥, 𝑦 = 𝐶𝑥
 
−1 1  
where 𝐴 = , 𝐶= 1 0
1 −2
Design a full-order state observer. The desired
observer poles are 𝑠 = −5 and 𝑠 = −5.

You might also like