Outline

• State variables.
State Variables • State-space representation.
• Linear state-space equations.
M. Sami Fadali • Nonlinear state-space equations.
Professor of Electrical Engineering • Linearization of state-space equations.
UNR

1 2

Input-output Description Linear Systems

• Single-input-single-output (SISO)
The description is valid for
a) time-varying systems: parameters are explicit functions
of time.
b) multi-input-multi-output (MIMO) systems: l input-output
differential equations, l = # of outputs.
c) nonlinear systems: differential equations include • Time-dependent coefficients for time-
nonlinear terms.
varying system.

3 4
 State Variables Definitions
To solve the differential equation we need System State: minimal set of numbers
(1) The system input for the period of interest. , needed together with
(2) A set of constant initial conditions. the input to uniquely
determine the behavior of the system in
the interval . = order of the system.
• Minimal set of initial conditions: incomplete
knowledge of the set prevents complete solution State Variables: As increases, the state of
but additional initial conditions are not needed to the system evolves and each of the
obtain the solution. numbers becomes a time variable.
• Initial conditions provide a summary of the history State Vector: vector of state variables
of the system up to the initial time.
5 6

Notation Definitions
• Column vector bolded State Space: -dimensional vector space where
represent the coordinate axes
• Row vector bolded and transposed . State plane: state space for a 2nd order system
Phase plane: special case where the state
variables are proportional to the derivatives of
.
Phase variables: state variables in phase plane.
State trajectories: Curves in state space
State portrait: plot of state trajectories in the plane
(phase portrait for the phase plane).

7 8
 Example 7.1 Example 7.1
• State for equation of motion of a point mass Write state-space equations for the spring-
driven by a force mass-damper system driven by a force
= displacement of the point mass. = mass = spring constant
= damper constant
= displacement of the point mass.

2  system is second order

9 10

Solution Solution (cont.)

• Obtain a complete solution given the • Two first order differential equations
force, if 2 initial conditions are known
: state of the system at time
.
• 2 I.C.s  system is second order.
• State (phase)Variables: [Link] equation: from definitions of state
variables.
• State Vector: [Link] equation: from equation of motion

11 12
 State-space Representation State Equations
• Set of first order equations governing the state
• Representation for the system described variables obtained from the input-output
by a differential equation in terms of state differential equation and the definitions of the state
variables.
and output equations.
• In general, n state equations for a nth order
• Linear Systems: More convenient to write system.
state (output) equations as a single matrix • The form of the state equations depends on the
equation. nature of the system (equations are time-varying
for time-varying systems, nonlinear for nonlinear
systems, etc.)
• State equations for linear time-invariant systems
can also be obtained from their transfer functions.
13 14

Output Equation Solution of State Equations

• Algebraic equation expressing the output • Solve the 1st order differential equations
in terms of the state variables and the then substitute in
input. • 2 differential equations + algebraic
• Multi-output systems: a scalar output expression are equivalent to the 2nd order
equation is needed to define each output. differential equation.
• Substitute from solution of state equation • Feedback Control Law 2nd order
to obtain output. underdamped system

15 16
 More on the Solution Phase Portrait
x2
[Link] depends only on initial conditions. 3

[Link] phase portrait using MATLAB 2

command lsim 1

[Link] is an implicit parameter.

0

[Link] indicate the direction of increasing

-1
time.
[Link] of state variables is not unique. -2

-3 x1
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2

17 18

Example 7.2: Matrix Form General Form for Linear Systems

Scalar Eqns. Matrix Form

real vectors.
Matrices:
state matrix
input or control matrix
output matrix
direct transmission matrix
19 20
 State Space in MATLAB Linear Vs. Nonlinear State-Space
» a = [0, 1; 5, 4];b=[0;1];c=[1,1];d=0;
» p = ss(a,b,c,d) % State-space quadruple Example 7.3: The following are examples of
a=
x1 x2
state-space equations for linear systems
x1 0 1
x2 -5 -4
a) 3rd order 2-input-2-output (MIMO) LTI
b=
u1  x1   11 . 0.3 15.   x1  01 . 0
 x    01      u1 
x1 0
. 35. 2.2 x 2  0 11 .  
x2 1    2
    u 2 
c=  x 3  0.4 2.4 11
.   x 3  10 . 
. 10
x1 x2
y1 1 1  x1 
 y1  1 2 0   1 2  u1 
 y   0 0 1  x 2   0 1 u 
d=
u1  2     2 
y1 0  x 3 
21 22
Continuous-time system.

Example 7.3 (b) Example 7.4: Nonlinear System

2nd order 2-output-1-input (SIMO) linear Obtain a state-space representation for
time-varying the s-D.O.F. robotic manipulator from the
 x1  sin(t ) cos(t )   x1  0 equations of motion with output q.
 x    1 
e 2 t   x 2  1
u
 2  D(q)q   C (q, q )q  g (q)  
 y1  1 0  x1  q  vector of generalized coordinates
 y   0 1  x  D(q)  s  s positive definite inertia matrix
 2   2 
C (q, q )  s  s matrix of velocity related terms
1. Zero D, constant B and C.
g (q)  s  1 vector of gravitational terms
2. Time-varying system: has entries that are functions of t.
  vector of generalized forces
23 24
 Solution Example 7.5
order 2 s (need 2 s initial conditions to solve completely.
State Variables Write the state-space equations for the 2-
x  colx1 , x 2   colq, q 
D.O.F. anthropomorphic manipulator.
State Equation
 x 1   x2   0  
x    D 1 (x )C (x , x )  g (x )   D 1 (x )u
 2  1 1 2 1   1 
 Link 2
u = generalized force
Output Equation y  x1
Link 1
x 
y  I s 0 ss  1 
x 2 
25 26

Equations of Motion Solution

m  m2 l12  m2l22  2m2l1l2 cos 2  m2l22  m2l1l2 cos 2  State vector
D    1 
 m2l22  m2l1l2 cos 2  m2l22 
 m l l sin  2  2 21   2 
  
C ,     2 1 2 
 m2l1l2 sin  2 12   x2 
x1  
m  m2 gl1 sin 1   m2 gl2 sin 1   2  
x   L m l l sin  2    m1  m2 gl1 sin1   m2 gl2 sin1 2 
  
2 1 2 2 2 1 2
g    1 
 2 
  m2l1l2 sin2 12   m2 gl2 sin1 2  
 m2 gl2 sin 1   2  
0
  u
mi li   masses (lengths) of two links.
 L

g  acceleration due to gravity. 1  m2l22 m2l22  m2l1l2 cos 2  

,    vectors of angular positions and velocities respectively. L 
det(D) m2l22  m2l1l2 cos 2 

m1  m2 l1  m2l2  2m2l1l2 cos 2 
2 2

27 28
 Nonlinear State-space Equations Linearization of State Equations
x  f ( x, u ) • Approximate nonlinear state equations by
y  g ( x, u ) linear state equations for small ranges of the
control and state variables.
f(.) (n1) and g(.) (l 1) = vectors of functions
• Linear equations are based on the first order
satisfying mathematical conditions to guarantee
approximation.
the existence and uniqueness of solution.
affine in the control: often encountered in
practice (includes equations of robotic
manipulators)
x  f (x) + B(x)u constant, = perturbation.
• Approximation Error of order
y  g(x) + D(x)u • Acceptable for small perturbations.
29 30

Function of Variables Nonlinear State-space Equations

• Approximation of the entry of the

vector

• Perturbation Vector
( , ) ( , )
measure of the length or "size" of the
perturbation vector (assumed negligible) ( , ) ( , )
Equilibrium:
31 32
Perturbations Abt’ Equilibrium
Derivative
( , ) ( , ) Equilibrium:

( , ) ( , )

( , ) ( , ) ( , )

( , ) ( , ) ( , )
33 34

Linearized State-Space
Output Equation
Equations
Approximation of entry of the vector  f1  f1    x1    f 1  f1    u1 
 ...  .   ...  . 
  x1  x 0 ,u 0   xn  x 0 ,u 0       u1  x 0 ,u 0   um x 0 ,u 0    
x   ... ... ...  . 
 ... ... ...  . 
 f  fn    f  fn  
 n ...  .   n ...  . 
  x1  x 0 ,u 0   x n  x0 ,u 0    x    u1
 n  x 0 ,u 0   um  x 0 ,u 0    u 
 m
( , ) ( , ) ( , )

  g1  g1    x1    g1  g1    u1 
 ...  .   ...  . 
  x1 x 0 , u 0   xn x 0 , u 0       u1 x 0 , u 0   un x 0 , u 0    
y   ... ... ...  .  ... ... ...  . 
( , ) ( , ) ( , )  g    g  
 gn  gn
 n ...  .   n ...  . 
  x1 x 0 , u 0 
 x n  x 0 , u 0    x    u1
n x 0 , u 0 
 u n  x 0 , u 0    u 
n

35 36
 Jacobians (drop ""s) Example 7.6
 f1  f1   f1  f1 
 ...   ...  Motion of nonlinear spring-mass-damper.
 x1 x 0 ,u 0 
 xn
x 0 ,u 0     u1 x 0 ,u 0 
 um
x 0 ,u 0  
A ... ... ...  B ... ... ... 
   f  = displacement = applied force
fn  fn  fn
 ...   n ... 
  x1  x 0 , u 0   x n  x 0 , u 0     u1 x 0 , u 0 
 um  x 0 , u 0   = mass of 1 Kg
= nonlinear damper constant
  g1  g1    g1  g1 
 ...   ...  = nonlinear spring force.
  x1 x 0 , u 0 
 xn
x 0 , u 0     u1 x 0 , u 0 
 um
x 0 , u 0  
C ... ... ...  D ... ... ...  (i) Find the equilibrium position
 g  gn   g  gn 
 n ...   n ...  corresponding to a force in terms of the
  x1 x 0 ,u 0 
 xn  x 0 , u 0     u1  0 0
x , u
 u m  0 0
x , u  spring constant, (ii) linearize the equation of
motion about this equilibrium.
37 38

Solution Linearize about the equilibrium

• Equilibrium with a force : Set all the time
derivatives equal to zero and solve for
• Equilibrium is at zero velocity and position .
= inverse function
• State Vector
• Input • Entries of state matrix: constants whose
values depend on the equilibrium.
• Nonlinear state equation
• Originally linear terms do not change with
linearization.
39 40
 Example MAPLE Example
• Nonlinear state equation  Jacobian( [x2,-2*(1+x1)*x2-4*x1^3+2*u],
[x1,x2] );

 Jacobian( [x2,-2*(1+x1)*x2-4*x1^3+2*u], [u] );

• Obtain Jacobian matrices using MAPLE

41 42