Académique Documents
Professionnel Documents
Culture Documents
1
FD CD AV 2
2
Where is the static air density, CD is Drag Coefficient and A is the projected
antenna area.
This equation is valid for both steady and time-varying wind gusts. The wind
velocity V is composed of a mean velocity Vm and a gust component Vg.
V Vm Vg
The projected area A is related to the array face area A 0 by the azimuth
angle of the array relative to wind and the antenna tilt back angle by
A A 0 cos(AZ ) cos(Tilt)
K cos(AZ ) V 2
(t ) K cos( t ) V V (t )
2
m g
x Ax Bu v
y Cx w
The disturbances (predominantly wind gusts) have covariance V. The
measurement noise has covariance W. It is assumed additionally that the
input and output noises are not correlated. This assumption is equivalent
to independence of their sources. Indeed, the measurement noise is
independent of the wind disturbances.
y Cxˆ y yˆ
The estimated state is obtained from the following equation:
u uf Flexible mode
x̂
+ Estimator
Controller
+
+
upI _
e
PI Controller +
Controller Modules
r +
GPS position and velocity are blended with the inertial data. Otherwise, if the
GPS data is not available, the system will operate without any GPS aiding.
The inertial navigator computes position, velocity and orientation of the IMU.
The Kalman filter estimates the errors in the inertial navigator along with IMU,
distance
GPS IMU
Kalman Filter
Position
Velocity
High long term and Acceleration
short term accuracy Angular Rate
Time
xk 1 k xk vk
zk H k xk wk
Error
GPS GPS Controller
Rx1 Rx2 Algorithm
Filter Correction Estimated Error
GPS Az
Measure Kalman Filter
Algorithm
GPS Observables
Difference between the position form the IMU and GPS is processed
In KF (typically @ 10Hz) to estimate the slowly growing position error
In the IMU.
External/Internal
Disturbance
Desired Actual
Azimuth Azimuth
Position Azimuth Antenna
Controller Drive System
IMU&GPS
Integrated
Algorithms
Kalman Filter
The estimation accuracy depends on the a priori information of the
system and measurement models, as well as the noise statistics.
A well-designed Kalman filter will attenuate the initial state errors, and
smooth the effects of system and measurement errors through the
averaging process.
Thank You
Dr. Ahmed Bassyouni IEEE Meeting-Syracuse 20