AEM 8421 / EE 5235: Exam #1 Thursday, March 22
1. Robust Stability: Consider the following set of models:
G(s)
I := : stable, kk < 1 (1)
1 + w(s)(s)
G(s) is SISO and w(s) is a stable, proper rational transfer function.
(a) Draw a block diagram showing the structure of any Gp I. Your diagram should only include blocks for
G, w, and .
(b) Consider the feedback diagram show in Figure 1. Assume the controller K stabilizes the nominal system
G. Find a necessary and sufficient condition for K to stabilize all Gp I. Provide a proof of sufficiency,
i.e. prove that K achieves robust stability if your condition is satisfied. You do not need to prove necessity.
r - e e - K(s) u - Gp (s) y -
6
Figure 1: Feedback Loop
2. Rear Steer Bicycle: In class we discussed a rear-steer bicycle designed by Richard Klein. Let denote the
steering angle of the rear wheel and the roll angle (rotation from vertical). The rider must choose to both
direct the bicycle and to stay upright ( = 0). The transfer function from input to output is:
V
s k2
G(s) = k1 V (2)
s2 hg
where g is the graviational constant, V is the bicycle speed, and h is the height of the center of mass. k1 and
k2 are constants that depend on the geometry of the bicycle.
(a) What are the locations of the open loop poles and zeros of the system G(s)?
(b) Derive bounds on the closed-loop bandwidth based on the locations of the open loop poles and zeros.
(c) Does the bicycle become easier or more difficult to control as V increases? Why?
(d) Suppose you re-design the bicycle to make it easier to control. Should you increase or decrease the height
h? Why?
3. Least Squares: Let A Cnm where n > m. The singular value decomposition (SVD) of A is given by:
A = U V = U1 U2 V
(3)
0nm,m
= diag(1 , . . . , m ). Assume A is full rank and hence i > 0 for
U and V are unitary matrices and
i = 1, . . . , m.
(a) What is the SVD of (A A)1 A ?
(b) Show that for any vectors x Cm and b Cn :
x U bk2 + kU bk2
kAx bk22 = kV (4)
1 2 2 2
(c) Use Equation 4 to specify the vector x that solves:
min kAx bk2 (5)
x
Note: You can complete this part even if you were not able to show the result in (b).
4. H Design: Consider the classical feedback loop shown in Figure 2 below. Assume that G(s) is the transfer
function for a SISO LTI system. The objective is to design a controller K(s) that stabilizes the system and
minimizes:
Wp S
Wu KS
(6)
S is the sensitivity function, Wp is the performance weight and Wu is the control effort weight. You may
assume the weights are proper, stable, rational transfer functions.
(a) Draw a block diagram of the classical feedback loop with additional blocks showing the locations of the
weights. Label all signals in your diagram.
(b) Draw a block diagram showing the generalized plant P for this design problem. Your diagram should
label the generalized disturbances, generalized errors, control measurements and control commands. In
other words, specify the generalized signals in terms of the signals shown in your diagram from part (a).
s
+
B
(c) Consider the weights Wu,1 (s) = 1 and Wp (s) = s+M
BA
. Sketch the Bode magnitude plot of Wp . Provide
an interpretation for the values M , B , and A in terms of the desired closed-loop performance.
(d) The design was performed using M = 1.5, B = 10 and A = 104 . The controller K1 designed with
these weights is shown in Figure 3. Suppose another controller K2 is designed with weights (Wp , Wu,2 )
s+10
where the new control weight is Wu,2 (s) = 0.1s+10 . Sketch the Bode magnitude plot of Wu,2 . How will a
controller K2 designed with weights (Wp , Wu,2 ) differ from the controller K1 ?
r -e
e
- K(s)
u
- G(s)
y
-
6
Figure 2: Feedback Loop
Figure 3: Bode magnitude plot for K1 .
