The objective of control systems is to maintain desired output with the best attainable
eciency, in which such performance of many modern technologies is enhanced
via minimizing or maximizing some state variables. Optimization problems are occasionally
handled under the condition such that extremum value, local plant gain and
required inputs are unknown. Sliding Mode Control (SMC) theory was developed
for the plants operate under uncertainty conditions and it exhibits high accuracy,
low sensitivity with respect to parameter variations and disturbances. SMC comply
with self optimization application requires controlling without gradient evaluation or
measurement. This is non-conventional control-system problem, cause gradient takes
part of the local gain and it can change sign. In the thesis, searching algorithm
is designed with monotonously decreasing or increasing function which is tracked
by the output of a plant until the extremum point is reached. Along with developed
design principle, aforementioned system behavior in the vicinity of extremum
is investigated through deeper analyses, cause the local gain of the plant tends to
zero and SMC inherently suers from the presence of high frequency oscillations, so
called chattering. This scope is considered as the theoretical part of the thesis and it
is complemented with comparison of other developed methodologies apply gradient
vector evaluation. Generalization of the proposed method is demonstrated for multidimensional
optimization. Beyond theoretical part, three self-optimization problems are under consideration, which are frequency control of DC/AC inverters implemented
on induction motors, output maximization of Underground Coal Gasication (UCG)
power plants and capability enhancement of a three-dimensional particle tracking
system by integration of an electrically focus-tunable lens.
The objective of minimizing the switching-frequency of DC/AC inverters with
desired width of hysteresis results in decreasing heat losses in the power converter.
Moreover, decreasing the width of hysteresis loop with remaining the level of admissible
frequency will increase the accuracy of inverters.
Gasication of coal deep below the earth surface, known as the Underground Coal
Gasication (UCG) process, is very sensitive to the
ow rate of oxidants. However,
the most critical factor is the steam to oxygen ratio in the input
ow rate. Sliding
modes based static self-optimization is applied to the UCG process which takes into
consideration the inlet pressure of the UCG reactor, adjusts the steam to oxygen
ratio such that the reactor output, which is the caloric value of the product gases,
is maximized. The simulation results are presented for the validated UCG model to
emphasize eectiveness of the algorithm.
SMC self-optimization model is used for tracking an axillary moving microscopic
particle of electrically focus tunable lens. Using lenses mounted on mechanical actuators
are not ecient against limited movement range, slower response times and
vibrations in the system. These drawbacks are often handled by using mathematical
models to estimate the state feedbacks before designing the control system. However,
the SM the self-optimizing design does not require state feedback information and
works by optimizing an output function of the system. The control system is then
successfully simulated. |