Lecture 11
Lecture 11
Lecture 11
b3
b2
b1
u(t) y(t)
FIR
t
Finite Impulse Response Model
One-Step-Ahead Predictor
[
yı(t θ ) = H −1 (q)G ( q)u(t ) + 1 − H −1 (q) y (t ) ]
= G ( q )u ( t )
yı (t θ ) = ϕ T (t )θ : linear regression
Given input data {u(t ), 1 ≤ t ≤ N } , the least square estimate of the parameter vector
was obtained as
N 2
1 1
θıNLS = arg min ∑ ( y(t ) − yı (t | θ )) = (R( N ))−1 f (N ) (40)
θ N t =1 2
where
N N
1 1
R(N ) =
N
∑ϕ (t )ϕ (t )
t =1
T
= ΦΦ T
and f ( N ) =
N
∑ ϕ (t ) y (t )
t =1
(41)
Cons
Two failure scenarios
• The impulse response has a slow decaying mode
1
• The sampling rate is high
The number of parameters, m, is too large to estimate.
The persistently exciting condition rankΦ = full rank can hardly be satisfied.
Check the eigenvalues of ΦΦ (or the singular values of Φ )
T
λ1 ≥ λ2 ≥ � ≥ λm
It is likely
λ1 ≥ λ2 ≥ � ≥ λn ≥ λn+1 ≅0 = � = λm
Often m becomes more than 50 and it is difficult to obtain such an input series having
50 non-zero singular values.
Time series data compression is an effective method for coping with this difficulty.
Before formulating the above least square estimate problem, data are processed so
that the information contained in the series of regressor may be represented in
compact, low-dimensional form.
ϕm
ϕn
.
. . . ϕ2 ϕ1
ϕ1 λ ≅0
.
i
Rm Rn n<m
L1(q) b1
u(t) y(t)
L2(q)
b2 ∑
�
Ln(q) bn
2
6.6 Continuous-Time Laguerre Series Expansion
[Theorem 6.6]
Assume a transfer function G(s) is
• Strictly proper lim G ( s ) = 0 G(∞) = 0 (45)
s →∞
• Continuous in Re(s) ≥ 0
k −1
∞
2a ⎛ s − a ⎞
G (s ) = ∑ g k ⎜ ⎟ (46)
k =1 s+a ⎝s+a⎠
where a > 0
Proof:
Consider the transformation given by
s+a
z= (47)
s−a
sz − az = s + a s ( z − 1) = a ( z + 1)
z +1
Bilinear transformation ∴s = a (48)
z −1
Im Im
s+a
s z = =1 for s = jω (42)
s−a z-plane s−a
1
∠z = ∠( s + a ) − ∠( s − a )
-1 1
Re Re
s
s−a s+a The imaginary axis is
mapped to the unit circle.
3
Im Im
-1 1
Re Re
⎛ ⎞
Consider ⎜ z +1⎟ (44) G (z ) is analytic outside
G (z) = G⎜ a ⎟
⎜ ��z −�1 ⎟
⎝ s ⎠ the unit circle.
∞
G ( z ) = ∑ g k z −k z >1 (49)
k =1
s+a
Considering that G (∞) = 0 lim = 1 ∴ G ( z ) = 0 at z=1
s →∞ s − a
∞
1
G ( z) = (1 − z −1 )∑ g k z −( k −1) (50)
2a k =1
s−a jω − a
Low Pass Filter
=1 All-pass Filter
s+a jω + a
4
a ≅pi , then a truncated Laguerre expansion:
k −1
n
2a ⎛ s − a ⎞
Gn (s) = ∑ g k ⎜ ⎟ → G( s ) (52)
k =1 s+a ⎝s+a⎠
converges to the original G(s) quickly.
- Recall -
For a continuous-time system, a pole close to the imaginary axis is slow to converge,
while a pole far from the imaginary axis converges quickly. Likewise, in discrete time,
a pole close to the origin of a z-plane quickly converges, while the ones near the unit
circle are slow.
s-plane Im z-plane
Choose a such that p1 ≅−a , where p1 is a slow, stable pole. Using the bilinear
transform, this slow pole in the s-plane can be transferred to a fast pole in the z-plane,
as shown below. Representing in the z-plane, the transfer function can be truncated;
just a few terms can approximate the impulse response since it converges quickly.
p1
Im
a ≅p1 Re
{g k }
Transformation
Im s+a
z=
1 s−a
Truncated
-1
time
1 Re
Fast Convergence
-1
Fast pole z ≅0
When the original transfer function has many poles, the Laguerre pole is placed near
the dominant pole so that most of the slow poles may be approximated by the
Laguerre pole.
5
Im Im
Dominant pole
1
-1 1
Re
z
Re
With the bilinear transform, these slow poles are transferred to the ones near the
origin of the z-plane. They are fast, hence the transfer function can be approximated
to a low order model.
[Theorem 6.7]
Assume that a Z-transform G(z) is
• Strictly proper G(∞) = 0
• Continuous in z ≥ 1
Then
k −1
∞
K ⎛ 1 − az ⎞
G (s ) = ∑ g k ⎜ ⎟ (53)
k =1 z−a⎝ z−a ⎠
where -1<a<1 and
K = (1 − a 2 )T (54)
Proof
Consider the bilinear transformation:
z−a
w= (55)
1 − az
w+a
w − azw = z − a w + a = z ( aw + 1) therefore, z = (56)
aw + 1
6
Transformation
z−a
z-plane w= w-plane
1 − az
⎛ w+a ⎞
G ( w) = G ⎜ ⎟
⎝ aw + 1 ⎠
z >1
w >1
⎛ w+a ⎞
G (w) = G ⎜ ⎟ is analytic in w > 1, and is proper G(∞) = 0
⎝ aw + 1 ⎠
1 1
lim w = − G (− ) = 0 (57)
z →∞ a a
Therefore
∞
T
G ( w) = (a + w −1 )∑ g k w −( k −1) (58)
K k =1
−(k −1)
⎛ z−a ⎞ T 1 − az ∞ ⎛ z−a ⎞
G( z ) = G ⎜ ⎟ = (a + )∑ g k ⎜ ⎟ (59)
⎝ 1 − az ⎠ K z − a k =1 ⎝ 1 − az ⎠
k −1
∞
K ⎛ 1 − az ⎞
∴ G( s) = ∑ g k ⎜ ⎟ (53)
k =1 z−a⎝ z−a ⎠
Q.E.D.
where Lk ( q ), k = 1,�, n is a series of filters. Once the original input data are filtered
7
with Lk ( q ), k = 1,�, n ,
x1 g1
L1(q)
y(t)
u(t) x2
L2(q) g2 ∑
xn
Ln (q ) gn
K Kq −1
x1 (t) = L1 ( q)u(t ) = u( t ) = u (t )
q−a 1 − aq −1
1 − aq q −1 − a
x2 (t) = x1 (t) = x1 (t)
q−a 1 − aq −1
x2 (t) − ax 2 (t − 1) = x1 (t − 1) − ax1 (t )
Recursive Filters