الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 183 |  |  |
| | | from | 183 | | |
Abstract
This thesis presents advanced control techniques for wind-generation systems. Two windgeneration
schemes are studied using different topologies (Line commutated converter,
controlled rectification and static reactive power compensator), different control strategies
(programmable high speed controller and pitch control) and different control structures
(PID-controller, Neural networks controller, Linear Quadratic Gaussian (LQG) and the Hinfinity
(H∞) controllers.
In the first wind generation scheme, a variable speed wind turbine driving a self-excited
induction generator, which is interfaced to the utility grid through asynchronous AC-DCAC
link. The variable amplitude, variable frequency voltage at the generator terminals is
first rectified and then the DC power is injected into the utility grid using a line
commutated inverter. The asynchronous DC-link virtually decouples the two systems,
allowing each to operate at its own frequency. The control objective aims to regulate the
rectifier output voltage and track the maximum available wind power. This is
accomplished via controlling the firing angles of the rectifier and the inverter. The
complete nonlinear dynamic model of the system has been described and linearized around
an operating point.
Two control structures have been employed. The first control is based on the linear
quadratic Gaussian approach, where a standard Kalman filter technique has been employed
to estimate the full states of the system. The computational burden has been minimized to a
great extent by computing the optimal state feedback gains and the Kalman state space
model off-line. The second control structure has been done based on H∞-synthesis to
control the DC link voltage and to track and extract maximum available wind power by
controlling the firing angles of the rectifier and the inverter. The design problem of the H∞-
controller has been formulated in a standard form with emphasis on the selection of the
weighting functions that reflect robustness and performance goals. The proposed system
has the advantages of robustness against model uncertainties and external disturbances, fast
response and the ability to reject noise.