Abstract
This dissertation presents some challenging problems in power system operations.
The efficacy of a heuristic method, namely, modified discrete particle swarm
optimization (MDPSO) algorithm is illustrated and compared with other methods by
solving the reliability based generator maintenance scheduling (GMS) optimization
problem of a practical hydrothermal power system. The concept of multiple swarms is
incorporated into the MDPSO algorithm to form a robust multiple swarms-modified
particle swarm optimization (MS-MDPSO) algorithm and applied to solving the GMS
problem on two power systems. Heuristic methods are proposed to circumvent the
problems of imposed non-smooth assumptions common with the classical approaches in
solving the challenging dynamic economic dispatch problem. The multi-objective
combined economic and emission dispatch (MO-CEED) optimization problem for a
wind-hydrothermal power system is formulated and solved in this dissertation. This MO-
CEED problem formulation becomes a challenging problem because of the presence of
uncertainty in wind power. A family of distributed optimal Pareto fronts for the MO-
CEED problem has been generated for different scenarios of capacity credit of wind
power. A real-time (RT) network stability index is formulated for determining a power
system’s ability to continue to provide service (electric energy) in a RT manner in case of
an unforeseen catastrophic contingency. Cascading stages of fuzzy inference system is
applied to combine non real-time (NRT) and RT power system assessments. NRT
analysis involves eigenvalue and transient energy analysis. RT analysis involves angle,
voltage and frequency stability indices. RT Network status index is implemented in real-
time on a practical power system.
