
27Control Strategies
where ω stands for random noise disturbance and ν represents random mea-
surement (sensor) noise. Both ω and ν are white Gaussian zero-mean station-
ary processes. It is noted that because the states and control are both random,
the performance index will be random. Thus, the problem is to nd the opti-
mal control that will minimize the average cost. Using the same procedure
as for the LQR problem, the solution is achieved as:
i. Controller:
0
1
1
u Kx
K R B P
A P PA PBR B P Q
T
T T
=
+ − + =
−
−
(2.19)
ii. Estimator:
( )
0
1 1
1
1 1 1 1
1
1 1
x Ax Bu K y Cx
K P C R
AP P A P C R CP Q
e
e
T
T T T
= + + −
=
+ − + Γ Γ =
−
−
(2.20)
In the previo ...